migrate.c 88.0 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0
2
/*
3
 * Memory Migration functionality - linux/mm/migrate.c
4 5 6 7 8 9 10 11 12
 *
 * Copyright (C) 2006 Silicon Graphics, Inc., Christoph Lameter
 *
 * Page migration was first developed in the context of the memory hotplug
 * project. The main authors of the migration code are:
 *
 * IWAMOTO Toshihiro <iwamoto@valinux.co.jp>
 * Hirokazu Takahashi <taka@valinux.co.jp>
 * Dave Hansen <haveblue@us.ibm.com>
C
Christoph Lameter 已提交
13
 * Christoph Lameter
14 15 16
 */

#include <linux/migrate.h>
17
#include <linux/export.h>
18
#include <linux/swap.h>
19
#include <linux/swapops.h>
20
#include <linux/pagemap.h>
21
#include <linux/buffer_head.h>
22
#include <linux/mm_inline.h>
23
#include <linux/nsproxy.h>
24
#include <linux/pagevec.h>
25
#include <linux/ksm.h>
26 27 28 29
#include <linux/rmap.h>
#include <linux/topology.h>
#include <linux/cpu.h>
#include <linux/cpuset.h>
30
#include <linux/writeback.h>
31 32
#include <linux/mempolicy.h>
#include <linux/vmalloc.h>
33
#include <linux/security.h>
34
#include <linux/backing-dev.h>
35
#include <linux/compaction.h>
36
#include <linux/syscalls.h>
37
#include <linux/compat.h>
38
#include <linux/hugetlb.h>
39
#include <linux/hugetlb_cgroup.h>
40
#include <linux/gfp.h>
41
#include <linux/pagewalk.h>
42
#include <linux/pfn_t.h>
43
#include <linux/memremap.h>
44
#include <linux/userfaultfd_k.h>
45
#include <linux/balloon_compaction.h>
46
#include <linux/mmu_notifier.h>
47
#include <linux/page_idle.h>
48
#include <linux/page_owner.h>
49
#include <linux/sched/mm.h>
50
#include <linux/ptrace.h>
51
#include <linux/oom.h>
52
#include <linux/memory.h>
53
#include <linux/random.h>
54

55 56
#include <asm/tlbflush.h>

57 58 59
#define CREATE_TRACE_POINTS
#include <trace/events/migrate.h>

60 61
#include "internal.h"

62
int isolate_movable_page(struct page *page, isolate_mode_t mode)
63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80
{
	struct address_space *mapping;

	/*
	 * Avoid burning cycles with pages that are yet under __free_pages(),
	 * or just got freed under us.
	 *
	 * In case we 'win' a race for a movable page being freed under us and
	 * raise its refcount preventing __free_pages() from doing its job
	 * the put_page() at the end of this block will take care of
	 * release this page, thus avoiding a nasty leakage.
	 */
	if (unlikely(!get_page_unless_zero(page)))
		goto out;

	/*
	 * Check PageMovable before holding a PG_lock because page's owner
	 * assumes anybody doesn't touch PG_lock of newly allocated page
81
	 * so unconditionally grabbing the lock ruins page's owner side.
82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112
	 */
	if (unlikely(!__PageMovable(page)))
		goto out_putpage;
	/*
	 * As movable pages are not isolated from LRU lists, concurrent
	 * compaction threads can race against page migration functions
	 * as well as race against the releasing a page.
	 *
	 * In order to avoid having an already isolated movable page
	 * being (wrongly) re-isolated while it is under migration,
	 * or to avoid attempting to isolate pages being released,
	 * lets be sure we have the page lock
	 * before proceeding with the movable page isolation steps.
	 */
	if (unlikely(!trylock_page(page)))
		goto out_putpage;

	if (!PageMovable(page) || PageIsolated(page))
		goto out_no_isolated;

	mapping = page_mapping(page);
	VM_BUG_ON_PAGE(!mapping, page);

	if (!mapping->a_ops->isolate_page(page, mode))
		goto out_no_isolated;

	/* Driver shouldn't use PG_isolated bit of page->flags */
	WARN_ON_ONCE(PageIsolated(page));
	__SetPageIsolated(page);
	unlock_page(page);

113
	return 0;
114 115 116 117 118 119

out_no_isolated:
	unlock_page(page);
out_putpage:
	put_page(page);
out:
120
	return -EBUSY;
121 122
}

123
static void putback_movable_page(struct page *page)
124 125 126 127 128 129 130 131
{
	struct address_space *mapping;

	mapping = page_mapping(page);
	mapping->a_ops->putback_page(page);
	__ClearPageIsolated(page);
}

132 133 134 135
/*
 * Put previously isolated pages back onto the appropriate lists
 * from where they were once taken off for compaction/migration.
 *
136 137 138
 * This function shall be used whenever the isolated pageset has been
 * built from lru, balloon, hugetlbfs page. See isolate_migratepages_range()
 * and isolate_huge_page().
139 140 141 142 143 144
 */
void putback_movable_pages(struct list_head *l)
{
	struct page *page;
	struct page *page2;

145
	list_for_each_entry_safe(page, page2, l, lru) {
146 147 148 149
		if (unlikely(PageHuge(page))) {
			putback_active_hugepage(page);
			continue;
		}
150
		list_del(&page->lru);
151 152 153 154 155
		/*
		 * We isolated non-lru movable page so here we can use
		 * __PageMovable because LRU page's mapping cannot have
		 * PAGE_MAPPING_MOVABLE.
		 */
156
		if (unlikely(__PageMovable(page))) {
157 158 159 160 161 162 163 164 165
			VM_BUG_ON_PAGE(!PageIsolated(page), page);
			lock_page(page);
			if (PageMovable(page))
				putback_movable_page(page);
			else
				__ClearPageIsolated(page);
			unlock_page(page);
			put_page(page);
		} else {
166
			mod_node_page_state(page_pgdat(page), NR_ISOLATED_ANON +
167
					page_is_file_lru(page), -thp_nr_pages(page));
168
			putback_lru_page(page);
169
		}
170 171 172
	}
}

173 174 175
/*
 * Restore a potential migration pte to a working pte entry
 */
176
static bool remove_migration_pte(struct page *page, struct vm_area_struct *vma,
177
				 unsigned long addr, void *old)
178
{
179 180 181 182 183 184 185 186
	struct page_vma_mapped_walk pvmw = {
		.page = old,
		.vma = vma,
		.address = addr,
		.flags = PVMW_SYNC | PVMW_MIGRATION,
	};
	struct page *new;
	pte_t pte;
187 188
	swp_entry_t entry;

189 190
	VM_BUG_ON_PAGE(PageTail(page), page);
	while (page_vma_mapped_walk(&pvmw)) {
191 192 193 194 195
		if (PageKsm(page))
			new = page;
		else
			new = page - pvmw.page->index +
				linear_page_index(vma, pvmw.address);
196

197 198 199 200 201 202 203 204 205
#ifdef CONFIG_ARCH_ENABLE_THP_MIGRATION
		/* PMD-mapped THP migration entry */
		if (!pvmw.pte) {
			VM_BUG_ON_PAGE(PageHuge(page) || !PageTransCompound(page), page);
			remove_migration_pmd(&pvmw, new);
			continue;
		}
#endif

206 207 208 209
		get_page(new);
		pte = pte_mkold(mk_pte(new, READ_ONCE(vma->vm_page_prot)));
		if (pte_swp_soft_dirty(*pvmw.pte))
			pte = pte_mksoft_dirty(pte);
210

211 212 213 214
		/*
		 * Recheck VMA as permissions can change since migration started
		 */
		entry = pte_to_swp_entry(*pvmw.pte);
215
		if (is_writable_migration_entry(entry))
216
			pte = maybe_mkwrite(pte, vma);
217 218
		else if (pte_swp_uffd_wp(*pvmw.pte))
			pte = pte_mkuffd_wp(pte);
219

220
		if (unlikely(is_device_private_page(new))) {
221 222 223 224 225 226
			if (pte_write(pte))
				entry = make_writable_device_private_entry(
							page_to_pfn(new));
			else
				entry = make_readable_device_private_entry(
							page_to_pfn(new));
227
			pte = swp_entry_to_pte(entry);
228 229
			if (pte_swp_soft_dirty(*pvmw.pte))
				pte = pte_swp_mksoft_dirty(pte);
230 231
			if (pte_swp_uffd_wp(*pvmw.pte))
				pte = pte_swp_mkuffd_wp(pte);
232
		}
233

234
#ifdef CONFIG_HUGETLB_PAGE
235
		if (PageHuge(new)) {
236 237
			unsigned int shift = huge_page_shift(hstate_vma(vma));

238
			pte = pte_mkhuge(pte);
239
			pte = arch_make_huge_pte(pte, shift, vma->vm_flags);
240 241 242 243
			if (PageAnon(new))
				hugepage_add_anon_rmap(new, vma, pvmw.address);
			else
				page_dup_rmap(new, true);
244
			set_huge_pte_at(vma->vm_mm, pvmw.address, pvmw.pte, pte);
245 246 247 248 249 250 251
		} else
#endif
		{
			if (PageAnon(new))
				page_add_anon_rmap(new, vma, pvmw.address, false);
			else
				page_add_file_rmap(new, false);
252
			set_pte_at(vma->vm_mm, pvmw.address, pvmw.pte, pte);
253
		}
254 255 256
		if (vma->vm_flags & VM_LOCKED && !PageTransCompound(new))
			mlock_vma_page(new);

257 258 259
		if (PageTransHuge(page) && PageMlocked(page))
			clear_page_mlock(page);

260 261 262
		/* No need to invalidate - it was non-present before */
		update_mmu_cache(vma, pvmw.address, pvmw.pte);
	}
263

264
	return true;
265 266
}

267 268 269 270
/*
 * Get rid of all migration entries and replace them by
 * references to the indicated page.
 */
271
void remove_migration_ptes(struct page *old, struct page *new, bool locked)
272
{
273 274 275 276 277
	struct rmap_walk_control rwc = {
		.rmap_one = remove_migration_pte,
		.arg = old,
	};

278 279 280 281
	if (locked)
		rmap_walk_locked(new, &rwc);
	else
		rmap_walk(new, &rwc);
282 283
}

284 285 286 287 288
/*
 * Something used the pte of a page under migration. We need to
 * get to the page and wait until migration is finished.
 * When we return from this function the fault will be retried.
 */
289
void __migration_entry_wait(struct mm_struct *mm, pte_t *ptep,
290
				spinlock_t *ptl)
291
{
292
	pte_t pte;
293 294
	swp_entry_t entry;

295
	spin_lock(ptl);
296 297 298 299 300 301 302 303
	pte = *ptep;
	if (!is_swap_pte(pte))
		goto out;

	entry = pte_to_swp_entry(pte);
	if (!is_migration_entry(entry))
		goto out;

304
	migration_entry_wait_on_locked(entry, ptep, ptl);
305 306 307 308 309
	return;
out:
	pte_unmap_unlock(ptep, ptl);
}

310 311 312 313 314 315 316 317
void migration_entry_wait(struct mm_struct *mm, pmd_t *pmd,
				unsigned long address)
{
	spinlock_t *ptl = pte_lockptr(mm, pmd);
	pte_t *ptep = pte_offset_map(pmd, address);
	__migration_entry_wait(mm, ptep, ptl);
}

318 319
void migration_entry_wait_huge(struct vm_area_struct *vma,
		struct mm_struct *mm, pte_t *pte)
320
{
321
	spinlock_t *ptl = huge_pte_lockptr(hstate_vma(vma), mm, pte);
322 323 324
	__migration_entry_wait(mm, pte, ptl);
}

325 326 327 328 329 330 331 332
#ifdef CONFIG_ARCH_ENABLE_THP_MIGRATION
void pmd_migration_entry_wait(struct mm_struct *mm, pmd_t *pmd)
{
	spinlock_t *ptl;

	ptl = pmd_lock(mm, pmd);
	if (!is_pmd_migration_entry(*pmd))
		goto unlock;
333
	migration_entry_wait_on_locked(pmd_to_swp_entry(*pmd), NULL, ptl);
334 335 336 337 338 339
	return;
unlock:
	spin_unlock(ptl);
}
#endif

340
static int expected_page_refs(struct address_space *mapping, struct page *page)
341 342 343 344
{
	int expected_count = 1;

	/*
345
	 * Device private pages have an extra refcount as they are
346 347 348
	 * ZONE_DEVICE pages.
	 */
	expected_count += is_device_private_page(page);
349
	if (mapping)
350
		expected_count += compound_nr(page) + page_has_private(page);
351 352 353 354

	return expected_count;
}

355
/*
356
 * Replace the page in the mapping.
357 358 359 360
 *
 * The number of remaining references must be:
 * 1 for anonymous pages without a mapping
 * 2 for pages with a mapping
361
 * 3 for pages with a mapping and PagePrivate/PagePrivate2 set.
362
 */
363 364
int folio_migrate_mapping(struct address_space *mapping,
		struct folio *newfolio, struct folio *folio, int extra_count)
365
{
366
	XA_STATE(xas, &mapping->i_pages, folio_index(folio));
367 368
	struct zone *oldzone, *newzone;
	int dirty;
369 370
	int expected_count = expected_page_refs(mapping, &folio->page) + extra_count;
	long nr = folio_nr_pages(folio);
371

372
	if (!mapping) {
373
		/* Anonymous page without mapping */
374
		if (folio_ref_count(folio) != expected_count)
375
			return -EAGAIN;
376 377

		/* No turning back from here */
378 379 380 381
		newfolio->index = folio->index;
		newfolio->mapping = folio->mapping;
		if (folio_test_swapbacked(folio))
			__folio_set_swapbacked(newfolio);
382

383
		return MIGRATEPAGE_SUCCESS;
384 385
	}

386 387
	oldzone = folio_zone(folio);
	newzone = folio_zone(newfolio);
388

389
	xas_lock_irq(&xas);
390
	if (!folio_ref_freeze(folio, expected_count)) {
391
		xas_unlock_irq(&xas);
392 393 394
		return -EAGAIN;
	}

395
	/*
396
	 * Now we know that no one else is looking at the folio:
397
	 * no turning back from here.
398
	 */
399 400 401 402 403 404 405 406
	newfolio->index = folio->index;
	newfolio->mapping = folio->mapping;
	folio_ref_add(newfolio, nr); /* add cache reference */
	if (folio_test_swapbacked(folio)) {
		__folio_set_swapbacked(newfolio);
		if (folio_test_swapcache(folio)) {
			folio_set_swapcache(newfolio);
			newfolio->private = folio_get_private(folio);
407 408
		}
	} else {
409
		VM_BUG_ON_FOLIO(folio_test_swapcache(folio), folio);
410 411
	}

412
	/* Move dirty while page refs frozen and newpage not yet exposed */
413
	dirty = folio_test_dirty(folio);
414
	if (dirty) {
415 416
		folio_clear_dirty(folio);
		folio_set_dirty(newfolio);
417 418
	}

419
	xas_store(&xas, newfolio);
420 421

	/*
422 423
	 * Drop cache reference from old page by unfreezing
	 * to one less reference.
424 425
	 * We know this isn't the last reference.
	 */
426
	folio_ref_unfreeze(folio, expected_count - nr);
427

428
	xas_unlock(&xas);
429 430
	/* Leave irq disabled to prevent preemption while updating stats */

431 432 433 434 435 436 437
	/*
	 * If moved to a different zone then also account
	 * the page for that zone. Other VM counters will be
	 * taken care of when we establish references to the
	 * new page and drop references to the old page.
	 *
	 * Note that anonymous pages are accounted for
438
	 * via NR_FILE_PAGES and NR_ANON_MAPPED if they
439 440
	 * are mapped to swap space.
	 */
441
	if (newzone != oldzone) {
442 443 444
		struct lruvec *old_lruvec, *new_lruvec;
		struct mem_cgroup *memcg;

445
		memcg = folio_memcg(folio);
446 447 448
		old_lruvec = mem_cgroup_lruvec(memcg, oldzone->zone_pgdat);
		new_lruvec = mem_cgroup_lruvec(memcg, newzone->zone_pgdat);

449 450
		__mod_lruvec_state(old_lruvec, NR_FILE_PAGES, -nr);
		__mod_lruvec_state(new_lruvec, NR_FILE_PAGES, nr);
451
		if (folio_test_swapbacked(folio) && !folio_test_swapcache(folio)) {
452 453
			__mod_lruvec_state(old_lruvec, NR_SHMEM, -nr);
			__mod_lruvec_state(new_lruvec, NR_SHMEM, nr);
454
		}
455
#ifdef CONFIG_SWAP
456
		if (folio_test_swapcache(folio)) {
457 458 459 460
			__mod_lruvec_state(old_lruvec, NR_SWAPCACHE, -nr);
			__mod_lruvec_state(new_lruvec, NR_SWAPCACHE, nr);
		}
#endif
461
		if (dirty && mapping_can_writeback(mapping)) {
462 463 464 465
			__mod_lruvec_state(old_lruvec, NR_FILE_DIRTY, -nr);
			__mod_zone_page_state(oldzone, NR_ZONE_WRITE_PENDING, -nr);
			__mod_lruvec_state(new_lruvec, NR_FILE_DIRTY, nr);
			__mod_zone_page_state(newzone, NR_ZONE_WRITE_PENDING, nr);
466
		}
467
	}
468
	local_irq_enable();
469

470
	return MIGRATEPAGE_SUCCESS;
471
}
472
EXPORT_SYMBOL(folio_migrate_mapping);
473

474 475
/*
 * The expected number of remaining references is the same as that
476
 * of folio_migrate_mapping().
477 478 479 480
 */
int migrate_huge_page_move_mapping(struct address_space *mapping,
				   struct page *newpage, struct page *page)
{
481
	XA_STATE(xas, &mapping->i_pages, page_index(page));
482 483
	int expected_count;

484
	xas_lock_irq(&xas);
485
	expected_count = 2 + page_has_private(page);
486 487
	if (page_count(page) != expected_count || xas_load(&xas) != page) {
		xas_unlock_irq(&xas);
488 489 490
		return -EAGAIN;
	}

491
	if (!page_ref_freeze(page, expected_count)) {
492
		xas_unlock_irq(&xas);
493 494 495
		return -EAGAIN;
	}

496 497
	newpage->index = page->index;
	newpage->mapping = page->mapping;
498

499 500
	get_page(newpage);

501
	xas_store(&xas, newpage);
502

503
	page_ref_unfreeze(page, expected_count - 1);
504

505
	xas_unlock_irq(&xas);
506

507
	return MIGRATEPAGE_SUCCESS;
508 509
}

510
/*
511
 * Copy the flags and some other ancillary information
512
 */
513
void folio_migrate_flags(struct folio *newfolio, struct folio *folio)
514
{
515 516
	int cpupid;

517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533
	if (folio_test_error(folio))
		folio_set_error(newfolio);
	if (folio_test_referenced(folio))
		folio_set_referenced(newfolio);
	if (folio_test_uptodate(folio))
		folio_mark_uptodate(newfolio);
	if (folio_test_clear_active(folio)) {
		VM_BUG_ON_FOLIO(folio_test_unevictable(folio), folio);
		folio_set_active(newfolio);
	} else if (folio_test_clear_unevictable(folio))
		folio_set_unevictable(newfolio);
	if (folio_test_workingset(folio))
		folio_set_workingset(newfolio);
	if (folio_test_checked(folio))
		folio_set_checked(newfolio);
	if (folio_test_mappedtodisk(folio))
		folio_set_mappedtodisk(newfolio);
534

535
	/* Move dirty on pages not done by folio_migrate_mapping() */
536 537
	if (folio_test_dirty(folio))
		folio_set_dirty(newfolio);
538

539 540 541 542
	if (folio_test_young(folio))
		folio_set_young(newfolio);
	if (folio_test_idle(folio))
		folio_set_idle(newfolio);
543

544 545 546 547
	/*
	 * Copy NUMA information to the new page, to prevent over-eager
	 * future migrations of this same page.
	 */
548 549
	cpupid = page_cpupid_xchg_last(&folio->page, -1);
	page_cpupid_xchg_last(&newfolio->page, cpupid);
550

551
	folio_migrate_ksm(newfolio, folio);
552 553 554 555
	/*
	 * Please do not reorder this without considering how mm/ksm.c's
	 * get_ksm_page() depends upon ksm_migrate_page() and PageSwapCache().
	 */
556 557 558
	if (folio_test_swapcache(folio))
		folio_clear_swapcache(folio);
	folio_clear_private(folio);
559 560

	/* page->private contains hugetlb specific flags */
561 562
	if (!folio_test_hugetlb(folio))
		folio->private = NULL;
563 564 565 566 567

	/*
	 * If any waiters have accumulated on the new page then
	 * wake them up.
	 */
568 569
	if (folio_test_writeback(newfolio))
		folio_end_writeback(newfolio);
570

571 572 573 574 575
	/*
	 * PG_readahead shares the same bit with PG_reclaim.  The above
	 * end_page_writeback() may clear PG_readahead mistakenly, so set the
	 * bit after that.
	 */
576 577
	if (folio_test_readahead(folio))
		folio_set_readahead(newfolio);
578

579
	folio_copy_owner(newfolio, folio);
580

581
	if (!folio_test_hugetlb(folio))
582
		mem_cgroup_migrate(folio, newfolio);
583
}
584
EXPORT_SYMBOL(folio_migrate_flags);
585

586
void folio_migrate_copy(struct folio *newfolio, struct folio *folio)
587
{
588 589
	folio_copy(newfolio, folio);
	folio_migrate_flags(newfolio, folio);
590
}
591
EXPORT_SYMBOL(folio_migrate_copy);
592

593 594 595 596
/************************************************************
 *                    Migration functions
 ***********************************************************/

597
/*
598
 * Common logic to directly migrate a single LRU page suitable for
599
 * pages that do not use PagePrivate/PagePrivate2.
600 601 602
 *
 * Pages are locked upon entry and exit.
 */
603
int migrate_page(struct address_space *mapping,
604 605
		struct page *newpage, struct page *page,
		enum migrate_mode mode)
606
{
607 608
	struct folio *newfolio = page_folio(newpage);
	struct folio *folio = page_folio(page);
609 610
	int rc;

611
	BUG_ON(folio_test_writeback(folio));	/* Writeback must be complete */
612

613
	rc = folio_migrate_mapping(mapping, newfolio, folio, 0);
614

615
	if (rc != MIGRATEPAGE_SUCCESS)
616 617
		return rc;

618
	if (mode != MIGRATE_SYNC_NO_COPY)
619
		folio_migrate_copy(newfolio, folio);
620
	else
621
		folio_migrate_flags(newfolio, folio);
622
	return MIGRATEPAGE_SUCCESS;
623 624 625
}
EXPORT_SYMBOL(migrate_page);

626
#ifdef CONFIG_BLOCK
627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664
/* Returns true if all buffers are successfully locked */
static bool buffer_migrate_lock_buffers(struct buffer_head *head,
							enum migrate_mode mode)
{
	struct buffer_head *bh = head;

	/* Simple case, sync compaction */
	if (mode != MIGRATE_ASYNC) {
		do {
			lock_buffer(bh);
			bh = bh->b_this_page;

		} while (bh != head);

		return true;
	}

	/* async case, we cannot block on lock_buffer so use trylock_buffer */
	do {
		if (!trylock_buffer(bh)) {
			/*
			 * We failed to lock the buffer and cannot stall in
			 * async migration. Release the taken locks
			 */
			struct buffer_head *failed_bh = bh;
			bh = head;
			while (bh != failed_bh) {
				unlock_buffer(bh);
				bh = bh->b_this_page;
			}
			return false;
		}

		bh = bh->b_this_page;
	} while (bh != head);
	return true;
}

665 666 667
static int __buffer_migrate_page(struct address_space *mapping,
		struct page *newpage, struct page *page, enum migrate_mode mode,
		bool check_refs)
668 669 670
{
	struct buffer_head *bh, *head;
	int rc;
671
	int expected_count;
672 673

	if (!page_has_buffers(page))
674
		return migrate_page(mapping, newpage, page, mode);
675

676
	/* Check whether page does not have extra refs before we do more work */
677
	expected_count = expected_page_refs(mapping, page);
678 679
	if (page_count(page) != expected_count)
		return -EAGAIN;
680

681 682 683
	head = page_buffers(page);
	if (!buffer_migrate_lock_buffers(head, mode))
		return -EAGAIN;
684

685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704
	if (check_refs) {
		bool busy;
		bool invalidated = false;

recheck_buffers:
		busy = false;
		spin_lock(&mapping->private_lock);
		bh = head;
		do {
			if (atomic_read(&bh->b_count)) {
				busy = true;
				break;
			}
			bh = bh->b_this_page;
		} while (bh != head);
		if (busy) {
			if (invalidated) {
				rc = -EAGAIN;
				goto unlock_buffers;
			}
705
			spin_unlock(&mapping->private_lock);
706 707 708 709 710 711
			invalidate_bh_lrus();
			invalidated = true;
			goto recheck_buffers;
		}
	}

712
	rc = migrate_page_move_mapping(mapping, newpage, page, 0);
713
	if (rc != MIGRATEPAGE_SUCCESS)
714
		goto unlock_buffers;
715

716
	attach_page_private(newpage, detach_page_private(page));
717 718 719 720 721 722 723 724

	bh = head;
	do {
		set_bh_page(bh, newpage, bh_offset(bh));
		bh = bh->b_this_page;

	} while (bh != head);

725 726 727 728
	if (mode != MIGRATE_SYNC_NO_COPY)
		migrate_page_copy(newpage, page);
	else
		migrate_page_states(newpage, page);
729

730 731
	rc = MIGRATEPAGE_SUCCESS;
unlock_buffers:
732 733
	if (check_refs)
		spin_unlock(&mapping->private_lock);
734 735 736 737 738 739 740
	bh = head;
	do {
		unlock_buffer(bh);
		bh = bh->b_this_page;

	} while (bh != head);

741
	return rc;
742
}
743 744 745 746 747 748 749 750 751 752 753

/*
 * Migration function for pages with buffers. This function can only be used
 * if the underlying filesystem guarantees that no other references to "page"
 * exist. For example attached buffer heads are accessed only under page lock.
 */
int buffer_migrate_page(struct address_space *mapping,
		struct page *newpage, struct page *page, enum migrate_mode mode)
{
	return __buffer_migrate_page(mapping, newpage, page, mode, false);
}
754
EXPORT_SYMBOL(buffer_migrate_page);
755 756 757 758 759 760 761 762 763 764 765 766

/*
 * Same as above except that this variant is more careful and checks that there
 * are also no buffer head references. This function is the right one for
 * mappings where buffer heads are directly looked up and referenced (such as
 * block device mappings).
 */
int buffer_migrate_page_norefs(struct address_space *mapping,
		struct page *newpage, struct page *page, enum migrate_mode mode)
{
	return __buffer_migrate_page(mapping, newpage, page, mode, true);
}
767
#endif
768

769 770 771 772
/*
 * Writeback a page to clean the dirty state
 */
static int writeout(struct address_space *mapping, struct page *page)
773
{
774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790
	struct writeback_control wbc = {
		.sync_mode = WB_SYNC_NONE,
		.nr_to_write = 1,
		.range_start = 0,
		.range_end = LLONG_MAX,
		.for_reclaim = 1
	};
	int rc;

	if (!mapping->a_ops->writepage)
		/* No write method for the address space */
		return -EINVAL;

	if (!clear_page_dirty_for_io(page))
		/* Someone else already triggered a write */
		return -EAGAIN;

791
	/*
792 793 794 795 796 797
	 * A dirty page may imply that the underlying filesystem has
	 * the page on some queue. So the page must be clean for
	 * migration. Writeout may mean we loose the lock and the
	 * page state is no longer what we checked for earlier.
	 * At this point we know that the migration attempt cannot
	 * be successful.
798
	 */
799
	remove_migration_ptes(page, page, false);
800

801
	rc = mapping->a_ops->writepage(page, &wbc);
802

803 804 805 806
	if (rc != AOP_WRITEPAGE_ACTIVATE)
		/* unlocked. Relock */
		lock_page(page);

807
	return (rc < 0) ? -EIO : -EAGAIN;
808 809 810 811 812 813
}

/*
 * Default handling if a filesystem does not provide a migration function.
 */
static int fallback_migrate_page(struct address_space *mapping,
814
	struct page *newpage, struct page *page, enum migrate_mode mode)
815
{
816
	if (PageDirty(page)) {
817
		/* Only writeback pages in full synchronous migration */
818 819 820 821 822
		switch (mode) {
		case MIGRATE_SYNC:
		case MIGRATE_SYNC_NO_COPY:
			break;
		default:
823
			return -EBUSY;
824
		}
825
		return writeout(mapping, page);
826
	}
827 828 829 830 831

	/*
	 * Buffers may be managed in a filesystem specific way.
	 * We must have no buffers or drop them.
	 */
832
	if (page_has_private(page) &&
833
	    !try_to_release_page(page, GFP_KERNEL))
834
		return mode == MIGRATE_SYNC ? -EAGAIN : -EBUSY;
835

836
	return migrate_page(mapping, newpage, page, mode);
837 838
}

839 840 841 842 843 844
/*
 * Move a page to a newly allocated page
 * The page is locked and all ptes have been successfully removed.
 *
 * The new page will have replaced the old page if this function
 * is successful.
845 846 847
 *
 * Return value:
 *   < 0 - error code
848
 *  MIGRATEPAGE_SUCCESS - success
849
 */
850
static int move_to_new_page(struct page *newpage, struct page *page,
851
				enum migrate_mode mode)
852 853
{
	struct address_space *mapping;
854 855
	int rc = -EAGAIN;
	bool is_lru = !__PageMovable(page);
856

857 858
	VM_BUG_ON_PAGE(!PageLocked(page), page);
	VM_BUG_ON_PAGE(!PageLocked(newpage), newpage);
859 860

	mapping = page_mapping(page);
861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878

	if (likely(is_lru)) {
		if (!mapping)
			rc = migrate_page(mapping, newpage, page, mode);
		else if (mapping->a_ops->migratepage)
			/*
			 * Most pages have a mapping and most filesystems
			 * provide a migratepage callback. Anonymous pages
			 * are part of swap space which also has its own
			 * migratepage callback. This is the most common path
			 * for page migration.
			 */
			rc = mapping->a_ops->migratepage(mapping, newpage,
							page, mode);
		else
			rc = fallback_migrate_page(mapping, newpage,
							page, mode);
	} else {
879
		/*
880 881
		 * In case of non-lru page, it could be released after
		 * isolation step. In that case, we shouldn't try migration.
882
		 */
883 884 885 886 887 888 889 890 891 892 893 894
		VM_BUG_ON_PAGE(!PageIsolated(page), page);
		if (!PageMovable(page)) {
			rc = MIGRATEPAGE_SUCCESS;
			__ClearPageIsolated(page);
			goto out;
		}

		rc = mapping->a_ops->migratepage(mapping, newpage,
						page, mode);
		WARN_ON_ONCE(rc == MIGRATEPAGE_SUCCESS &&
			!PageIsolated(page));
	}
895

896 897 898 899 900
	/*
	 * When successful, old pagecache page->mapping must be cleared before
	 * page is freed; but stats require that PageAnon be left as PageAnon.
	 */
	if (rc == MIGRATEPAGE_SUCCESS) {
901 902 903 904 905 906 907 908 909 910 911
		if (__PageMovable(page)) {
			VM_BUG_ON_PAGE(!PageIsolated(page), page);

			/*
			 * We clear PG_movable under page_lock so any compactor
			 * cannot try to migrate this page.
			 */
			__ClearPageIsolated(page);
		}

		/*
912
		 * Anonymous and movable page->mapping will be cleared by
913 914 915 916
		 * free_pages_prepare so don't reset it here for keeping
		 * the type to work PageAnon, for example.
		 */
		if (!PageMappingFlags(page))
917
			page->mapping = NULL;
918

919 920
		if (likely(!is_zone_device_page(newpage)))
			flush_dcache_folio(page_folio(newpage));
921
	}
922
out:
923 924 925
	return rc;
}

926
static int __unmap_and_move(struct page *page, struct page *newpage,
927
				int force, enum migrate_mode mode)
928
{
929
	int rc = -EAGAIN;
930
	bool page_was_mapped = false;
931
	struct anon_vma *anon_vma = NULL;
932
	bool is_lru = !__PageMovable(page);
933

N
Nick Piggin 已提交
934
	if (!trylock_page(page)) {
935
		if (!force || mode == MIGRATE_ASYNC)
936
			goto out;
937 938 939 940 941 942 943

		/*
		 * It's not safe for direct compaction to call lock_page.
		 * For example, during page readahead pages are added locked
		 * to the LRU. Later, when the IO completes the pages are
		 * marked uptodate and unlocked. However, the queueing
		 * could be merging multiple pages for one bio (e.g.
944
		 * mpage_readahead). If an allocation happens for the
945 946 947 948 949 950 951
		 * second or third page, the process can end up locking
		 * the same page twice and deadlocking. Rather than
		 * trying to be clever about what pages can be locked,
		 * avoid the use of lock_page for direct compaction
		 * altogether.
		 */
		if (current->flags & PF_MEMALLOC)
952
			goto out;
953

954 955 956 957
		lock_page(page);
	}

	if (PageWriteback(page)) {
958
		/*
959
		 * Only in the case of a full synchronous migration is it
960 961 962
		 * necessary to wait for PageWriteback. In the async case,
		 * the retry loop is too short and in the sync-light case,
		 * the overhead of stalling is too much
963
		 */
964 965 966 967 968
		switch (mode) {
		case MIGRATE_SYNC:
		case MIGRATE_SYNC_NO_COPY:
			break;
		default:
969
			rc = -EBUSY;
970
			goto out_unlock;
971 972
		}
		if (!force)
973
			goto out_unlock;
974 975
		wait_on_page_writeback(page);
	}
976

977
	/*
978
	 * By try_to_migrate(), page->mapcount goes down to 0 here. In this case,
979
	 * we cannot notice that anon_vma is freed while we migrates a page.
980
	 * This get_anon_vma() delays freeing anon_vma pointer until the end
981
	 * of migration. File cache pages are no problem because of page_lock()
982 983
	 * File Caches may use write_page() or lock_page() in migration, then,
	 * just care Anon page here.
984 985 986 987 988 989
	 *
	 * Only page_get_anon_vma() understands the subtleties of
	 * getting a hold on an anon_vma from outside one of its mms.
	 * But if we cannot get anon_vma, then we won't need it anyway,
	 * because that implies that the anon page is no longer mapped
	 * (and cannot be remapped so long as we hold the page lock).
990
	 */
991
	if (PageAnon(page) && !PageKsm(page))
992
		anon_vma = page_get_anon_vma(page);
993

994 995 996 997 998 999 1000 1001 1002 1003 1004
	/*
	 * Block others from accessing the new page when we get around to
	 * establishing additional references. We are usually the only one
	 * holding a reference to newpage at this point. We used to have a BUG
	 * here if trylock_page(newpage) fails, but would like to allow for
	 * cases where there might be a race with the previous use of newpage.
	 * This is much like races on refcount of oldpage: just don't BUG().
	 */
	if (unlikely(!trylock_page(newpage)))
		goto out_unlock;

1005 1006 1007 1008 1009
	if (unlikely(!is_lru)) {
		rc = move_to_new_page(newpage, page, mode);
		goto out_unlock_both;
	}

1010
	/*
1011 1012 1013 1014 1015
	 * Corner case handling:
	 * 1. When a new swap-cache page is read into, it is added to the LRU
	 * and treated as swapcache but it has no rmap yet.
	 * Calling try_to_unmap() against a page->mapping==NULL page will
	 * trigger a BUG.  So handle it here.
1016
	 * 2. An orphaned page (see truncate_cleanup_page) might have
1017 1018 1019 1020
	 * fs-private metadata. The page can be picked up due to memory
	 * offlining.  Everywhere else except page reclaim, the page is
	 * invisible to the vm, so the page can not be migrated.  So try to
	 * free the metadata, so the page can be freed.
1021
	 */
1022
	if (!page->mapping) {
1023
		VM_BUG_ON_PAGE(PageAnon(page), page);
1024
		if (page_has_private(page)) {
1025
			try_to_free_buffers(page);
1026
			goto out_unlock_both;
1027
		}
1028 1029
	} else if (page_mapped(page)) {
		/* Establish migration ptes */
1030 1031
		VM_BUG_ON_PAGE(PageAnon(page) && !PageKsm(page) && !anon_vma,
				page);
1032
		try_to_migrate(page, 0);
1033
		page_was_mapped = true;
1034
	}
1035

1036
	if (!page_mapped(page))
1037
		rc = move_to_new_page(newpage, page, mode);
1038

1039 1040
	if (page_was_mapped)
		remove_migration_ptes(page,
1041
			rc == MIGRATEPAGE_SUCCESS ? newpage : page, false);
1042

1043 1044 1045
out_unlock_both:
	unlock_page(newpage);
out_unlock:
1046
	/* Drop an anon_vma reference if we took one */
1047
	if (anon_vma)
1048
		put_anon_vma(anon_vma);
1049
	unlock_page(page);
1050
out:
1051 1052 1053 1054
	/*
	 * If migration is successful, decrease refcount of the newpage
	 * which will not free the page because new page owner increased
	 * refcounter. As well, if it is LRU page, add the page to LRU
1055 1056 1057 1058
	 * list in here. Use the old state of the isolated source page to
	 * determine if we migrated a LRU page. newpage was already unlocked
	 * and possibly modified by its owner - don't rely on the page
	 * state.
1059 1060
	 */
	if (rc == MIGRATEPAGE_SUCCESS) {
1061
		if (unlikely(!is_lru))
1062 1063 1064 1065 1066
			put_page(newpage);
		else
			putback_lru_page(newpage);
	}

1067 1068
	return rc;
}
1069

1070 1071 1072 1073
/*
 * Obtain the lock on page, remove all ptes and migrate the page
 * to the newly allocated page in newpage.
 */
1074
static int unmap_and_move(new_page_t get_new_page,
1075 1076
				   free_page_t put_new_page,
				   unsigned long private, struct page *page,
1077
				   int force, enum migrate_mode mode,
1078 1079
				   enum migrate_reason reason,
				   struct list_head *ret)
1080
{
1081
	int rc = MIGRATEPAGE_SUCCESS;
1082
	struct page *newpage = NULL;
1083

1084
	if (!thp_migration_supported() && PageTransHuge(page))
1085
		return -ENOSYS;
1086

1087 1088
	if (page_count(page) == 1) {
		/* page was freed from under us. So we are done. */
1089 1090
		ClearPageActive(page);
		ClearPageUnevictable(page);
1091 1092 1093 1094 1095 1096
		if (unlikely(__PageMovable(page))) {
			lock_page(page);
			if (!PageMovable(page))
				__ClearPageIsolated(page);
			unlock_page(page);
		}
1097 1098 1099
		goto out;
	}

1100 1101 1102 1103
	newpage = get_new_page(page, private);
	if (!newpage)
		return -ENOMEM;

1104
	rc = __unmap_and_move(page, newpage, force, mode);
1105
	if (rc == MIGRATEPAGE_SUCCESS)
1106
		set_page_owner_migrate_reason(newpage, reason);
1107

1108
out:
1109
	if (rc != -EAGAIN) {
1110 1111 1112
		/*
		 * A page that has been migrated has all references
		 * removed and will be freed. A page that has not been
1113
		 * migrated will have kept its references and be restored.
1114 1115
		 */
		list_del(&page->lru);
1116
	}
1117

1118 1119 1120 1121 1122 1123
	/*
	 * If migration is successful, releases reference grabbed during
	 * isolation. Otherwise, restore the page to right list unless
	 * we want to retry.
	 */
	if (rc == MIGRATEPAGE_SUCCESS) {
1124 1125 1126 1127 1128 1129
		/*
		 * Compaction can migrate also non-LRU pages which are
		 * not accounted to NR_ISOLATED_*. They can be recognized
		 * as __PageMovable
		 */
		if (likely(!__PageMovable(page)))
1130
			mod_node_page_state(page_pgdat(page), NR_ISOLATED_ANON +
1131
					page_is_file_lru(page), -thp_nr_pages(page));
1132

1133
		if (reason != MR_MEMORY_FAILURE)
1134
			/*
1135
			 * We release the page in page_handle_poison.
1136
			 */
1137
			put_page(page);
1138
	} else {
1139 1140
		if (rc != -EAGAIN)
			list_add_tail(&page->lru, ret);
1141

1142 1143 1144 1145
		if (put_new_page)
			put_new_page(newpage, private);
		else
			put_page(newpage);
1146
	}
1147

1148 1149 1150
	return rc;
}

1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169
/*
 * Counterpart of unmap_and_move_page() for hugepage migration.
 *
 * This function doesn't wait the completion of hugepage I/O
 * because there is no race between I/O and migration for hugepage.
 * Note that currently hugepage I/O occurs only in direct I/O
 * where no lock is held and PG_writeback is irrelevant,
 * and writeback status of all subpages are counted in the reference
 * count of the head page (i.e. if all subpages of a 2MB hugepage are
 * under direct I/O, the reference of the head page is 512 and a bit more.)
 * This means that when we try to migrate hugepage whose subpages are
 * doing direct I/O, some references remain after try_to_unmap() and
 * hugepage migration fails without data corruption.
 *
 * There is also no race when direct I/O is issued on the page under migration,
 * because then pte is replaced with migration swap entry and direct I/O code
 * will wait in the page fault for migration to complete.
 */
static int unmap_and_move_huge_page(new_page_t get_new_page,
1170 1171
				free_page_t put_new_page, unsigned long private,
				struct page *hpage, int force,
1172 1173
				enum migrate_mode mode, int reason,
				struct list_head *ret)
1174
{
1175
	int rc = -EAGAIN;
1176
	int page_was_mapped = 0;
1177
	struct page *new_hpage;
1178
	struct anon_vma *anon_vma = NULL;
1179
	struct address_space *mapping = NULL;
1180

1181
	/*
1182
	 * Migratability of hugepages depends on architectures and their size.
1183 1184 1185 1186 1187
	 * This check is necessary because some callers of hugepage migration
	 * like soft offline and memory hotremove don't walk through page
	 * tables or check whether the hugepage is pmd-based or not before
	 * kicking migration.
	 */
1188
	if (!hugepage_migration_supported(page_hstate(hpage))) {
1189
		list_move_tail(&hpage->lru, ret);
1190
		return -ENOSYS;
1191
	}
1192

1193 1194 1195 1196 1197 1198
	if (page_count(hpage) == 1) {
		/* page was freed from under us. So we are done. */
		putback_active_hugepage(hpage);
		return MIGRATEPAGE_SUCCESS;
	}

1199
	new_hpage = get_new_page(hpage, private);
1200 1201 1202 1203
	if (!new_hpage)
		return -ENOMEM;

	if (!trylock_page(hpage)) {
1204
		if (!force)
1205
			goto out;
1206 1207 1208 1209 1210 1211 1212
		switch (mode) {
		case MIGRATE_SYNC:
		case MIGRATE_SYNC_NO_COPY:
			break;
		default:
			goto out;
		}
1213 1214 1215
		lock_page(hpage);
	}

1216 1217 1218 1219 1220
	/*
	 * Check for pages which are in the process of being freed.  Without
	 * page_mapping() set, hugetlbfs specific move page routine will not
	 * be called and we could leak usage counts for subpools.
	 */
1221
	if (hugetlb_page_subpool(hpage) && !page_mapping(hpage)) {
1222 1223 1224 1225
		rc = -EBUSY;
		goto out_unlock;
	}

1226 1227
	if (PageAnon(hpage))
		anon_vma = page_get_anon_vma(hpage);
1228

1229 1230 1231
	if (unlikely(!trylock_page(new_hpage)))
		goto put_anon;

1232
	if (page_mapped(hpage)) {
1233
		bool mapping_locked = false;
1234
		enum ttu_flags ttu = 0;
1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249

		if (!PageAnon(hpage)) {
			/*
			 * In shared mappings, try_to_unmap could potentially
			 * call huge_pmd_unshare.  Because of this, take
			 * semaphore in write mode here and set TTU_RMAP_LOCKED
			 * to let lower levels know we have taken the lock.
			 */
			mapping = hugetlb_page_mapping_lock_write(hpage);
			if (unlikely(!mapping))
				goto unlock_put_anon;

			mapping_locked = true;
			ttu |= TTU_RMAP_LOCKED;
		}
1250

1251
		try_to_migrate(hpage, ttu);
1252
		page_was_mapped = 1;
1253 1254 1255

		if (mapping_locked)
			i_mmap_unlock_write(mapping);
1256
	}
1257 1258

	if (!page_mapped(hpage))
1259
		rc = move_to_new_page(new_hpage, hpage, mode);
1260

1261
	if (page_was_mapped)
1262
		remove_migration_ptes(hpage,
1263
			rc == MIGRATEPAGE_SUCCESS ? new_hpage : hpage, false);
1264

1265
unlock_put_anon:
1266 1267 1268
	unlock_page(new_hpage);

put_anon:
H
Hugh Dickins 已提交
1269
	if (anon_vma)
1270
		put_anon_vma(anon_vma);
1271

1272
	if (rc == MIGRATEPAGE_SUCCESS) {
1273
		move_hugetlb_state(hpage, new_hpage, reason);
1274 1275
		put_new_page = NULL;
	}
1276

1277
out_unlock:
1278
	unlock_page(hpage);
1279
out:
1280
	if (rc == MIGRATEPAGE_SUCCESS)
1281
		putback_active_hugepage(hpage);
1282
	else if (rc != -EAGAIN)
1283
		list_move_tail(&hpage->lru, ret);
1284 1285 1286 1287 1288 1289

	/*
	 * If migration was not successful and there's a freeing callback, use
	 * it.  Otherwise, put_page() will drop the reference grabbed during
	 * isolation.
	 */
1290
	if (put_new_page)
1291 1292
		put_new_page(new_hpage, private);
	else
1293
		putback_active_hugepage(new_hpage);
1294

1295 1296 1297
	return rc;
}

1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311
static inline int try_split_thp(struct page *page, struct page **page2,
				struct list_head *from)
{
	int rc = 0;

	lock_page(page);
	rc = split_huge_page_to_list(page, from);
	unlock_page(page);
	if (!rc)
		list_safe_reset_next(page, *page2, lru);

	return rc;
}

1312
/*
1313 1314
 * migrate_pages - migrate the pages specified in a list, to the free pages
 *		   supplied as the target for the page migration
1315
 *
1316 1317 1318
 * @from:		The list of pages to be migrated.
 * @get_new_page:	The function used to allocate free pages to be used
 *			as the target of the page migration.
1319 1320
 * @put_new_page:	The function used to free target pages if migration
 *			fails, or NULL if no special handling is necessary.
1321 1322 1323 1324
 * @private:		Private data to be passed on to get_new_page()
 * @mode:		The migration mode that specifies the constraints for
 *			page migration, if any.
 * @reason:		The reason for page migration.
1325
 * @ret_succeeded:	Set to the number of normal pages migrated successfully if
1326
 *			the caller passes a non-NULL pointer.
1327
 *
1328 1329
 * The function returns after 10 attempts or if no pages are movable any more
 * because the list has become empty or no retryable pages exist any more.
1330 1331
 * It is caller's responsibility to call putback_movable_pages() to return pages
 * to the LRU or free list only if ret != 0.
1332
 *
1333 1334 1335
 * Returns the number of {normal page, THP, hugetlb} that were not migrated, or
 * an error code. The number of THP splits will be considered as the number of
 * non-migrated THP, no matter how many subpages of the THP are migrated successfully.
1336
 */
1337
int migrate_pages(struct list_head *from, new_page_t get_new_page,
1338
		free_page_t put_new_page, unsigned long private,
1339
		enum migrate_mode mode, int reason, unsigned int *ret_succeeded)
1340
{
1341
	int retry = 1;
1342
	int thp_retry = 1;
1343
	int nr_failed = 0;
1344
	int nr_failed_pages = 0;
1345
	int nr_succeeded = 0;
1346 1347 1348
	int nr_thp_succeeded = 0;
	int nr_thp_failed = 0;
	int nr_thp_split = 0;
1349
	int pass = 0;
1350
	bool is_thp = false;
1351 1352 1353
	struct page *page;
	struct page *page2;
	int swapwrite = current->flags & PF_SWAPWRITE;
1354
	int rc, nr_subpages;
1355
	LIST_HEAD(ret_pages);
1356
	LIST_HEAD(thp_split_pages);
1357
	bool nosplit = (reason == MR_NUMA_MISPLACED);
1358
	bool no_subpage_counting = false;
1359

1360 1361
	trace_mm_migrate_pages_start(mode, reason);

1362 1363 1364
	if (!swapwrite)
		current->flags |= PF_SWAPWRITE;

1365
thp_subpage_migration:
1366
	for (pass = 0; pass < 10 && (retry || thp_retry); pass++) {
1367
		retry = 0;
1368
		thp_retry = 0;
1369

1370
		list_for_each_entry_safe(page, page2, from, lru) {
1371
retry:
1372 1373 1374 1375 1376
			/*
			 * THP statistics is based on the source huge page.
			 * Capture required information that might get lost
			 * during migration.
			 */
1377
			is_thp = PageTransHuge(page) && !PageHuge(page);
1378
			nr_subpages = compound_nr(page);
1379
			cond_resched();
1380

1381 1382
			if (PageHuge(page))
				rc = unmap_and_move_huge_page(get_new_page,
1383
						put_new_page, private, page,
1384 1385
						pass > 2, mode, reason,
						&ret_pages);
1386
			else
1387
				rc = unmap_and_move(get_new_page, put_new_page,
1388
						private, page, pass > 2, mode,
1389 1390 1391 1392 1393 1394 1395 1396 1397 1398
						reason, &ret_pages);
			/*
			 * The rules are:
			 *	Success: non hugetlb page will be freed, hugetlb
			 *		 page will be put back
			 *	-EAGAIN: stay on the from list
			 *	-ENOMEM: stay on the from list
			 *	Other errno: put on ret_pages list then splice to
			 *		     from list
			 */
1399
			switch(rc) {
1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413
			/*
			 * THP migration might be unsupported or the
			 * allocation could've failed so we should
			 * retry on the same page with the THP split
			 * to base pages.
			 *
			 * Head page is retried immediately and tail
			 * pages are added to the tail of the list so
			 * we encounter them after the rest of the list
			 * is processed.
			 */
			case -ENOSYS:
				/* THP migration is unsupported */
				if (is_thp) {
1414 1415
					nr_thp_failed++;
					if (!try_split_thp(page, &page2, &thp_split_pages)) {
1416 1417 1418 1419
						nr_thp_split++;
						goto retry;
					}

1420
					nr_failed_pages += nr_subpages;
1421 1422 1423 1424
					break;
				}

				/* Hugetlb migration is unsupported */
1425 1426
				if (!no_subpage_counting)
					nr_failed++;
1427
				nr_failed_pages += nr_subpages;
1428
				break;
1429
			case -ENOMEM:
1430
				/*
1431 1432
				 * When memory is low, don't bother to try to migrate
				 * other pages, just exit.
1433
				 * THP NUMA faulting doesn't split THP to retry.
1434
				 */
1435
				if (is_thp && !nosplit) {
1436 1437
					nr_thp_failed++;
					if (!try_split_thp(page, &page2, &thp_split_pages)) {
1438
						nr_thp_split++;
1439 1440
						goto retry;
					}
1441

1442
					nr_failed_pages += nr_subpages;
1443 1444
					goto out;
				}
1445 1446 1447

				if (!no_subpage_counting)
					nr_failed++;
1448
				nr_failed_pages += nr_subpages;
1449
				goto out;
1450
			case -EAGAIN:
1451 1452 1453 1454
				if (is_thp) {
					thp_retry++;
					break;
				}
1455
				retry++;
1456
				break;
1457
			case MIGRATEPAGE_SUCCESS:
1458
				nr_succeeded += nr_subpages;
1459 1460 1461 1462
				if (is_thp) {
					nr_thp_succeeded++;
					break;
				}
1463 1464
				break;
			default:
1465
				/*
1466
				 * Permanent failure (-EBUSY, etc.):
1467 1468 1469 1470
				 * unlike -EAGAIN case, the failed page is
				 * removed from migration page list and not
				 * retried in the next outer loop.
				 */
1471 1472
				if (is_thp) {
					nr_thp_failed++;
1473
					nr_failed_pages += nr_subpages;
1474 1475
					break;
				}
1476 1477 1478

				if (!no_subpage_counting)
					nr_failed++;
1479
				nr_failed_pages += nr_subpages;
1480
				break;
1481
			}
1482 1483
		}
	}
1484
	nr_failed += retry;
1485
	nr_thp_failed += thp_retry;
1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503
	/*
	 * Try to migrate subpages of fail-to-migrate THPs, no nr_failed
	 * counting in this round, since all subpages of a THP is counted
	 * as 1 failure in the first round.
	 */
	if (!list_empty(&thp_split_pages)) {
		/*
		 * Move non-migrated pages (after 10 retries) to ret_pages
		 * to avoid migrating them again.
		 */
		list_splice_init(from, &ret_pages);
		list_splice_init(&thp_split_pages, from);
		no_subpage_counting = true;
		retry = 1;
		goto thp_subpage_migration;
	}

	rc = nr_failed + nr_thp_failed;
1504
out:
1505 1506 1507 1508 1509 1510
	/*
	 * Put the permanent failure page back to migration list, they
	 * will be put back to the right list by the caller.
	 */
	list_splice(&ret_pages, from);

1511
	count_vm_events(PGMIGRATE_SUCCESS, nr_succeeded);
1512
	count_vm_events(PGMIGRATE_FAIL, nr_failed_pages);
1513 1514 1515
	count_vm_events(THP_MIGRATION_SUCCESS, nr_thp_succeeded);
	count_vm_events(THP_MIGRATION_FAIL, nr_thp_failed);
	count_vm_events(THP_MIGRATION_SPLIT, nr_thp_split);
1516
	trace_mm_migrate_pages(nr_succeeded, nr_failed_pages, nr_thp_succeeded,
1517
			       nr_thp_failed, nr_thp_split, mode, reason);
1518

1519 1520 1521
	if (!swapwrite)
		current->flags &= ~PF_SWAPWRITE;

1522 1523 1524
	if (ret_succeeded)
		*ret_succeeded = nr_succeeded;

1525
	return rc;
1526
}
1527

1528
struct page *alloc_migration_target(struct page *page, unsigned long private)
1529
{
1530 1531
	struct migration_target_control *mtc;
	gfp_t gfp_mask;
1532 1533
	unsigned int order = 0;
	struct page *new_page = NULL;
1534 1535 1536 1537 1538 1539 1540 1541
	int nid;
	int zidx;

	mtc = (struct migration_target_control *)private;
	gfp_mask = mtc->gfp_mask;
	nid = mtc->nid;
	if (nid == NUMA_NO_NODE)
		nid = page_to_nid(page);
1542

1543 1544 1545
	if (PageHuge(page)) {
		struct hstate *h = page_hstate(compound_head(page));

1546 1547
		gfp_mask = htlb_modify_alloc_mask(h, gfp_mask);
		return alloc_huge_page_nodemask(h, nid, mtc->nmask, gfp_mask);
1548
	}
1549 1550

	if (PageTransHuge(page)) {
1551 1552 1553 1554 1555
		/*
		 * clear __GFP_RECLAIM to make the migration callback
		 * consistent with regular THP allocations.
		 */
		gfp_mask &= ~__GFP_RECLAIM;
1556 1557 1558
		gfp_mask |= GFP_TRANSHUGE;
		order = HPAGE_PMD_ORDER;
	}
1559 1560
	zidx = zone_idx(page_zone(page));
	if (is_highmem_idx(zidx) || zidx == ZONE_MOVABLE)
1561 1562
		gfp_mask |= __GFP_HIGHMEM;

1563
	new_page = __alloc_pages(gfp_mask, order, nid, mtc->nmask);
1564 1565 1566 1567 1568 1569 1570

	if (new_page && PageTransHuge(new_page))
		prep_transhuge_page(new_page);

	return new_page;
}

1571 1572
#ifdef CONFIG_NUMA

1573
static int store_status(int __user *status, int start, int value, int nr)
1574
{
1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587
	while (nr-- > 0) {
		if (put_user(value, status + start))
			return -EFAULT;
		start++;
	}

	return 0;
}

static int do_move_pages_to_node(struct mm_struct *mm,
		struct list_head *pagelist, int node)
{
	int err;
1588 1589 1590 1591
	struct migration_target_control mtc = {
		.nid = node,
		.gfp_mask = GFP_HIGHUSER_MOVABLE | __GFP_THISNODE,
	};
1592

1593
	err = migrate_pages(pagelist, alloc_migration_target, NULL,
1594
		(unsigned long)&mtc, MIGRATE_SYNC, MR_SYSCALL, NULL);
1595 1596 1597
	if (err)
		putback_movable_pages(pagelist);
	return err;
1598 1599 1600
}

/*
1601 1602
 * Resolves the given address to a struct page, isolates it from the LRU and
 * puts it to the given pagelist.
1603 1604 1605 1606 1607
 * Returns:
 *     errno - if the page cannot be found/isolated
 *     0 - when it doesn't have to be migrated because it is already on the
 *         target node
 *     1 - when it has been queued
1608
 */
1609 1610
static int add_page_for_migration(struct mm_struct *mm, unsigned long addr,
		int node, struct list_head *pagelist, bool migrate_all)
1611
{
1612 1613 1614
	struct vm_area_struct *vma;
	struct page *page;
	unsigned int follflags;
1615 1616
	int err;

1617
	mmap_read_lock(mm);
1618 1619 1620 1621
	err = -EFAULT;
	vma = find_vma(mm, addr);
	if (!vma || addr < vma->vm_start || !vma_migratable(vma))
		goto out;
1622

1623 1624 1625
	/* FOLL_DUMP to ignore special (like zero) pages */
	follflags = FOLL_GET | FOLL_DUMP;
	page = follow_page(vma, addr, follflags);
1626

1627 1628 1629
	err = PTR_ERR(page);
	if (IS_ERR(page))
		goto out;
1630

1631 1632 1633
	err = -ENOENT;
	if (!page)
		goto out;
1634

1635 1636 1637
	err = 0;
	if (page_to_nid(page) == node)
		goto out_putpage;
1638

1639 1640 1641
	err = -EACCES;
	if (page_mapcount(page) > 1 && !migrate_all)
		goto out_putpage;
1642

1643 1644 1645
	if (PageHuge(page)) {
		if (PageHead(page)) {
			isolate_huge_page(page, pagelist);
1646
			err = 1;
1647
		}
1648 1649
	} else {
		struct page *head;
1650

1651 1652
		head = compound_head(page);
		err = isolate_lru_page(head);
1653
		if (err)
1654
			goto out_putpage;
1655

1656
		err = 1;
1657 1658
		list_add_tail(&head->lru, pagelist);
		mod_node_page_state(page_pgdat(head),
1659
			NR_ISOLATED_ANON + page_is_file_lru(head),
1660
			thp_nr_pages(head));
1661 1662 1663 1664 1665 1666 1667 1668 1669
	}
out_putpage:
	/*
	 * Either remove the duplicate refcount from
	 * isolate_lru_page() or drop the page ref if it was
	 * not isolated.
	 */
	put_page(page);
out:
1670
	mmap_read_unlock(mm);
1671 1672 1673
	return err;
}

1674 1675 1676 1677 1678 1679
static int move_pages_and_store_status(struct mm_struct *mm, int node,
		struct list_head *pagelist, int __user *status,
		int start, int i, unsigned long nr_pages)
{
	int err;

1680 1681 1682
	if (list_empty(pagelist))
		return 0;

1683 1684 1685 1686 1687 1688
	err = do_move_pages_to_node(mm, pagelist, node);
	if (err) {
		/*
		 * Positive err means the number of failed
		 * pages to migrate.  Since we are going to
		 * abort and return the number of non-migrated
1689
		 * pages, so need to include the rest of the
1690 1691 1692 1693 1694 1695 1696 1697 1698 1699
		 * nr_pages that have not been attempted as
		 * well.
		 */
		if (err > 0)
			err += nr_pages - i - 1;
		return err;
	}
	return store_status(status, start, node, i - start);
}

1700 1701 1702 1703
/*
 * Migrate an array of page address onto an array of nodes and fill
 * the corresponding array of status.
 */
1704
static int do_pages_move(struct mm_struct *mm, nodemask_t task_nodes,
1705 1706 1707 1708 1709
			 unsigned long nr_pages,
			 const void __user * __user *pages,
			 const int __user *nodes,
			 int __user *status, int flags)
{
1710 1711 1712 1713
	int current_node = NUMA_NO_NODE;
	LIST_HEAD(pagelist);
	int start, i;
	int err = 0, err1;
1714

1715
	lru_cache_disable();
1716

1717 1718 1719 1720
	for (i = start = 0; i < nr_pages; i++) {
		const void __user *p;
		unsigned long addr;
		int node;
1721

1722 1723 1724 1725 1726
		err = -EFAULT;
		if (get_user(p, pages + i))
			goto out_flush;
		if (get_user(node, nodes + i))
			goto out_flush;
1727
		addr = (unsigned long)untagged_addr(p);
1728 1729 1730 1731 1732 1733

		err = -ENODEV;
		if (node < 0 || node >= MAX_NUMNODES)
			goto out_flush;
		if (!node_state(node, N_MEMORY))
			goto out_flush;
1734

1735 1736 1737 1738 1739 1740 1741 1742
		err = -EACCES;
		if (!node_isset(node, task_nodes))
			goto out_flush;

		if (current_node == NUMA_NO_NODE) {
			current_node = node;
			start = i;
		} else if (node != current_node) {
1743 1744
			err = move_pages_and_store_status(mm, current_node,
					&pagelist, status, start, i, nr_pages);
1745 1746 1747 1748
			if (err)
				goto out;
			start = i;
			current_node = node;
1749 1750
		}

1751 1752 1753 1754 1755 1756
		/*
		 * Errors in the page lookup or isolation are not fatal and we simply
		 * report them via status
		 */
		err = add_page_for_migration(mm, addr, current_node,
				&pagelist, flags & MPOL_MF_MOVE_ALL);
1757

1758
		if (err > 0) {
1759 1760 1761
			/* The page is successfully queued for migration */
			continue;
		}
1762

1763 1764 1765 1766 1767 1768 1769
		/*
		 * The move_pages() man page does not have an -EEXIST choice, so
		 * use -EFAULT instead.
		 */
		if (err == -EEXIST)
			err = -EFAULT;

1770 1771 1772 1773 1774
		/*
		 * If the page is already on the target node (!err), store the
		 * node, otherwise, store the err.
		 */
		err = store_status(status, i, err ? : current_node, 1);
1775 1776
		if (err)
			goto out_flush;
1777

1778 1779
		err = move_pages_and_store_status(mm, current_node, &pagelist,
				status, start, i, nr_pages);
1780 1781
		if (err)
			goto out;
1782
		current_node = NUMA_NO_NODE;
1783
	}
1784 1785
out_flush:
	/* Make sure we do not overwrite the existing error */
1786 1787
	err1 = move_pages_and_store_status(mm, current_node, &pagelist,
				status, start, i, nr_pages);
1788
	if (err >= 0)
1789
		err = err1;
1790
out:
1791
	lru_cache_enable();
1792 1793 1794
	return err;
}

1795
/*
1796
 * Determine the nodes of an array of pages and store it in an array of status.
1797
 */
1798 1799
static void do_pages_stat_array(struct mm_struct *mm, unsigned long nr_pages,
				const void __user **pages, int *status)
1800
{
1801 1802
	unsigned long i;

1803
	mmap_read_lock(mm);
1804

1805
	for (i = 0; i < nr_pages; i++) {
1806
		unsigned long addr = (unsigned long)(*pages);
1807 1808
		struct vm_area_struct *vma;
		struct page *page;
1809
		int err = -EFAULT;
1810

1811 1812
		vma = vma_lookup(mm, addr);
		if (!vma)
1813 1814
			goto set_status;

1815 1816
		/* FOLL_DUMP to ignore special (like zero) pages */
		page = follow_page(vma, addr, FOLL_DUMP);
1817 1818 1819 1820 1821

		err = PTR_ERR(page);
		if (IS_ERR(page))
			goto set_status;

1822
		err = page ? page_to_nid(page) : -ENOENT;
1823
set_status:
1824 1825 1826 1827 1828 1829
		*status = err;

		pages++;
		status++;
	}

1830
	mmap_read_unlock(mm);
1831 1832
}

1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849
static int get_compat_pages_array(const void __user *chunk_pages[],
				  const void __user * __user *pages,
				  unsigned long chunk_nr)
{
	compat_uptr_t __user *pages32 = (compat_uptr_t __user *)pages;
	compat_uptr_t p;
	int i;

	for (i = 0; i < chunk_nr; i++) {
		if (get_user(p, pages32 + i))
			return -EFAULT;
		chunk_pages[i] = compat_ptr(p);
	}

	return 0;
}

1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861
/*
 * Determine the nodes of a user array of pages and store it in
 * a user array of status.
 */
static int do_pages_stat(struct mm_struct *mm, unsigned long nr_pages,
			 const void __user * __user *pages,
			 int __user *status)
{
#define DO_PAGES_STAT_CHUNK_NR 16
	const void __user *chunk_pages[DO_PAGES_STAT_CHUNK_NR];
	int chunk_status[DO_PAGES_STAT_CHUNK_NR];

1862 1863
	while (nr_pages) {
		unsigned long chunk_nr;
1864

1865 1866 1867 1868
		chunk_nr = nr_pages;
		if (chunk_nr > DO_PAGES_STAT_CHUNK_NR)
			chunk_nr = DO_PAGES_STAT_CHUNK_NR;

1869 1870 1871 1872 1873 1874 1875 1876 1877
		if (in_compat_syscall()) {
			if (get_compat_pages_array(chunk_pages, pages,
						   chunk_nr))
				break;
		} else {
			if (copy_from_user(chunk_pages, pages,
				      chunk_nr * sizeof(*chunk_pages)))
				break;
		}
1878 1879 1880

		do_pages_stat_array(mm, chunk_nr, chunk_pages, chunk_status);

1881 1882
		if (copy_to_user(status, chunk_status, chunk_nr * sizeof(*status)))
			break;
1883

1884 1885 1886 1887 1888
		pages += chunk_nr;
		status += chunk_nr;
		nr_pages -= chunk_nr;
	}
	return nr_pages ? -EFAULT : 0;
1889 1890
}

1891
static struct mm_struct *find_mm_struct(pid_t pid, nodemask_t *mem_nodes)
1892 1893 1894 1895
{
	struct task_struct *task;
	struct mm_struct *mm;

1896 1897 1898 1899 1900 1901 1902 1903 1904
	/*
	 * There is no need to check if current process has the right to modify
	 * the specified process when they are same.
	 */
	if (!pid) {
		mmget(current->mm);
		*mem_nodes = cpuset_mems_allowed(current);
		return current->mm;
	}
1905 1906

	/* Find the mm_struct */
1907
	rcu_read_lock();
1908
	task = find_task_by_vpid(pid);
1909
	if (!task) {
1910
		rcu_read_unlock();
1911
		return ERR_PTR(-ESRCH);
1912
	}
1913
	get_task_struct(task);
1914 1915 1916

	/*
	 * Check if this process has the right to modify the specified
1917
	 * process. Use the regular "ptrace_may_access()" checks.
1918
	 */
1919
	if (!ptrace_may_access(task, PTRACE_MODE_READ_REALCREDS)) {
1920
		rcu_read_unlock();
1921
		mm = ERR_PTR(-EPERM);
1922
		goto out;
1923
	}
1924
	rcu_read_unlock();
1925

1926 1927
	mm = ERR_PTR(security_task_movememory(task));
	if (IS_ERR(mm))
1928
		goto out;
1929
	*mem_nodes = cpuset_mems_allowed(task);
1930
	mm = get_task_mm(task);
1931
out:
1932
	put_task_struct(task);
1933
	if (!mm)
1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952
		mm = ERR_PTR(-EINVAL);
	return mm;
}

/*
 * Move a list of pages in the address space of the currently executing
 * process.
 */
static int kernel_move_pages(pid_t pid, unsigned long nr_pages,
			     const void __user * __user *pages,
			     const int __user *nodes,
			     int __user *status, int flags)
{
	struct mm_struct *mm;
	int err;
	nodemask_t task_nodes;

	/* Check flags */
	if (flags & ~(MPOL_MF_MOVE|MPOL_MF_MOVE_ALL))
1953 1954
		return -EINVAL;

1955 1956 1957 1958 1959 1960 1961
	if ((flags & MPOL_MF_MOVE_ALL) && !capable(CAP_SYS_NICE))
		return -EPERM;

	mm = find_mm_struct(pid, &task_nodes);
	if (IS_ERR(mm))
		return PTR_ERR(mm);

1962 1963 1964 1965 1966
	if (nodes)
		err = do_pages_move(mm, task_nodes, nr_pages, pages,
				    nodes, status, flags);
	else
		err = do_pages_stat(mm, nr_pages, pages, status);
1967 1968 1969 1970 1971

	mmput(mm);
	return err;
}

1972 1973 1974 1975 1976 1977 1978 1979
SYSCALL_DEFINE6(move_pages, pid_t, pid, unsigned long, nr_pages,
		const void __user * __user *, pages,
		const int __user *, nodes,
		int __user *, status, int, flags)
{
	return kernel_move_pages(pid, nr_pages, pages, nodes, status, flags);
}

1980 1981 1982 1983 1984 1985
#ifdef CONFIG_NUMA_BALANCING
/*
 * Returns true if this is a safe migration target node for misplaced NUMA
 * pages. Currently it only checks the watermarks which crude
 */
static bool migrate_balanced_pgdat(struct pglist_data *pgdat,
1986
				   unsigned long nr_migrate_pages)
1987 1988
{
	int z;
1989

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
	for (z = pgdat->nr_zones - 1; z >= 0; z--) {
		struct zone *zone = pgdat->node_zones + z;

		if (!populated_zone(zone))
			continue;

		/* Avoid waking kswapd by allocating pages_to_migrate pages. */
		if (!zone_watermark_ok(zone, 0,
				       high_wmark_pages(zone) +
				       nr_migrate_pages,
2000
				       ZONE_MOVABLE, 0))
2001 2002 2003 2004 2005 2006 2007
			continue;
		return true;
	}
	return false;
}

static struct page *alloc_misplaced_dst_page(struct page *page,
2008
					   unsigned long data)
2009 2010 2011 2012
{
	int nid = (int) data;
	struct page *newpage;

2013
	newpage = __alloc_pages_node(nid,
2014 2015 2016
					 (GFP_HIGHUSER_MOVABLE |
					  __GFP_THISNODE | __GFP_NOMEMALLOC |
					  __GFP_NORETRY | __GFP_NOWARN) &
2017
					 ~__GFP_RECLAIM, 0);
2018

2019 2020 2021
	return newpage;
}

2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038
static struct page *alloc_misplaced_dst_page_thp(struct page *page,
						 unsigned long data)
{
	int nid = (int) data;
	struct page *newpage;

	newpage = alloc_pages_node(nid, (GFP_TRANSHUGE_LIGHT | __GFP_THISNODE),
				   HPAGE_PMD_ORDER);
	if (!newpage)
		goto out;

	prep_transhuge_page(newpage);

out:
	return newpage;
}

2039
static int numamigrate_isolate_page(pg_data_t *pgdat, struct page *page)
2040
{
2041
	int page_lru;
2042
	int nr_pages = thp_nr_pages(page);
2043

2044
	VM_BUG_ON_PAGE(compound_order(page) && !PageTransHuge(page), page);
2045

2046 2047 2048 2049
	/* Do not migrate THP mapped by multiple processes */
	if (PageTransHuge(page) && total_mapcount(page) > 1)
		return 0;

2050
	/* Avoid migrating to a node that is nearly full */
2051
	if (!migrate_balanced_pgdat(pgdat, nr_pages))
2052
		return 0;
2053

2054 2055
	if (isolate_lru_page(page))
		return 0;
2056

2057
	page_lru = page_is_file_lru(page);
2058
	mod_node_page_state(page_pgdat(page), NR_ISOLATED_ANON + page_lru,
2059
			    nr_pages);
2060

2061
	/*
2062 2063 2064
	 * Isolating the page has taken another reference, so the
	 * caller's reference can be safely dropped without the page
	 * disappearing underneath us during migration.
2065 2066
	 */
	put_page(page);
2067
	return 1;
2068 2069 2070 2071 2072 2073 2074
}

/*
 * Attempt to migrate a misplaced page to the specified destination
 * node. Caller is expected to have an elevated reference count on
 * the page that will be dropped by this function before returning.
 */
2075 2076
int migrate_misplaced_page(struct page *page, struct vm_area_struct *vma,
			   int node)
2077 2078
{
	pg_data_t *pgdat = NODE_DATA(node);
2079
	int isolated;
2080 2081
	int nr_remaining;
	LIST_HEAD(migratepages);
2082 2083
	new_page_t *new;
	bool compound;
2084
	int nr_pages = thp_nr_pages(page);
2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096

	/*
	 * PTE mapped THP or HugeTLB page can't reach here so the page could
	 * be either base page or THP.  And it must be head page if it is
	 * THP.
	 */
	compound = PageTransHuge(page);

	if (compound)
		new = alloc_misplaced_dst_page_thp;
	else
		new = alloc_misplaced_dst_page;
2097 2098

	/*
2099 2100
	 * Don't migrate file pages that are mapped in multiple processes
	 * with execute permissions as they are probably shared libraries.
2101
	 */
2102 2103
	if (page_mapcount(page) != 1 && page_is_file_lru(page) &&
	    (vma->vm_flags & VM_EXEC))
2104 2105
		goto out;

2106 2107 2108 2109
	/*
	 * Also do not migrate dirty pages as not all filesystems can move
	 * dirty pages in MIGRATE_ASYNC mode which is a waste of cycles.
	 */
2110
	if (page_is_file_lru(page) && PageDirty(page))
2111 2112
		goto out;

2113 2114 2115 2116 2117
	isolated = numamigrate_isolate_page(pgdat, page);
	if (!isolated)
		goto out;

	list_add(&page->lru, &migratepages);
2118
	nr_remaining = migrate_pages(&migratepages, *new, NULL, node,
2119
				     MIGRATE_ASYNC, MR_NUMA_MISPLACED, NULL);
2120
	if (nr_remaining) {
2121 2122
		if (!list_empty(&migratepages)) {
			list_del(&page->lru);
2123 2124
			mod_node_page_state(page_pgdat(page), NR_ISOLATED_ANON +
					page_is_file_lru(page), -nr_pages);
2125 2126
			putback_lru_page(page);
		}
2127 2128
		isolated = 0;
	} else
2129
		count_vm_numa_events(NUMA_PAGE_MIGRATE, nr_pages);
2130 2131
	BUG_ON(!list_empty(&migratepages));
	return isolated;
2132 2133 2134 2135

out:
	put_page(page);
	return 0;
2136
}
2137
#endif /* CONFIG_NUMA_BALANCING */
2138
#endif /* CONFIG_NUMA */
2139

2140
#ifdef CONFIG_DEVICE_PRIVATE
2141
static int migrate_vma_collect_skip(unsigned long start,
2142 2143 2144 2145 2146 2147
				    unsigned long end,
				    struct mm_walk *walk)
{
	struct migrate_vma *migrate = walk->private;
	unsigned long addr;

2148
	for (addr = start; addr < end; addr += PAGE_SIZE) {
2149
		migrate->dst[migrate->npages] = 0;
2150
		migrate->src[migrate->npages++] = 0;
2151 2152 2153 2154 2155
	}

	return 0;
}

2156
static int migrate_vma_collect_hole(unsigned long start,
2157
				    unsigned long end,
2158
				    __always_unused int depth,
2159 2160 2161 2162 2163
				    struct mm_walk *walk)
{
	struct migrate_vma *migrate = walk->private;
	unsigned long addr;

2164 2165 2166 2167
	/* Only allow populating anonymous memory. */
	if (!vma_is_anonymous(walk->vma))
		return migrate_vma_collect_skip(start, end, walk);

2168
	for (addr = start; addr < end; addr += PAGE_SIZE) {
2169
		migrate->src[migrate->npages] = MIGRATE_PFN_MIGRATE;
2170
		migrate->dst[migrate->npages] = 0;
2171 2172
		migrate->npages++;
		migrate->cpages++;
2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185
	}

	return 0;
}

static int migrate_vma_collect_pmd(pmd_t *pmdp,
				   unsigned long start,
				   unsigned long end,
				   struct mm_walk *walk)
{
	struct migrate_vma *migrate = walk->private;
	struct vm_area_struct *vma = walk->vma;
	struct mm_struct *mm = vma->vm_mm;
2186
	unsigned long addr = start, unmapped = 0;
2187 2188 2189 2190 2191
	spinlock_t *ptl;
	pte_t *ptep;

again:
	if (pmd_none(*pmdp))
2192
		return migrate_vma_collect_hole(start, end, -1, walk);
2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207

	if (pmd_trans_huge(*pmdp)) {
		struct page *page;

		ptl = pmd_lock(mm, pmdp);
		if (unlikely(!pmd_trans_huge(*pmdp))) {
			spin_unlock(ptl);
			goto again;
		}

		page = pmd_page(*pmdp);
		if (is_huge_zero_page(page)) {
			spin_unlock(ptl);
			split_huge_pmd(vma, pmdp, addr);
			if (pmd_trans_unstable(pmdp))
2208
				return migrate_vma_collect_skip(start, end,
2209 2210 2211 2212 2213 2214 2215
								walk);
		} else {
			int ret;

			get_page(page);
			spin_unlock(ptl);
			if (unlikely(!trylock_page(page)))
2216
				return migrate_vma_collect_skip(start, end,
2217 2218 2219 2220
								walk);
			ret = split_huge_page(page);
			unlock_page(page);
			put_page(page);
2221 2222 2223 2224
			if (ret)
				return migrate_vma_collect_skip(start, end,
								walk);
			if (pmd_none(*pmdp))
2225
				return migrate_vma_collect_hole(start, end, -1,
2226 2227 2228 2229 2230
								walk);
		}
	}

	if (unlikely(pmd_bad(*pmdp)))
2231
		return migrate_vma_collect_skip(start, end, walk);
2232 2233

	ptep = pte_offset_map_lock(mm, pmdp, addr, &ptl);
2234 2235
	arch_enter_lazy_mmu_mode();

2236
	for (; addr < end; addr += PAGE_SIZE, ptep++) {
2237
		unsigned long mpfn = 0, pfn;
2238
		struct page *page;
2239
		swp_entry_t entry;
2240 2241 2242 2243
		pte_t pte;

		pte = *ptep;

2244
		if (pte_none(pte)) {
2245 2246 2247 2248
			if (vma_is_anonymous(vma)) {
				mpfn = MIGRATE_PFN_MIGRATE;
				migrate->cpages++;
			}
2249 2250 2251
			goto next;
		}

2252 2253 2254 2255 2256 2257 2258 2259 2260 2261
		if (!pte_present(pte)) {
			/*
			 * Only care about unaddressable device page special
			 * page table entry. Other special swap entries are not
			 * migratable, and we ignore regular swapped page.
			 */
			entry = pte_to_swp_entry(pte);
			if (!is_device_private_entry(entry))
				goto next;

2262
			page = pfn_swap_entry_to_page(entry);
2263 2264 2265
			if (!(migrate->flags &
				MIGRATE_VMA_SELECT_DEVICE_PRIVATE) ||
			    page->pgmap->owner != migrate->pgmap_owner)
2266 2267
				goto next;

2268 2269
			mpfn = migrate_pfn(page_to_pfn(page)) |
					MIGRATE_PFN_MIGRATE;
2270
			if (is_writable_device_private_entry(entry))
2271 2272
				mpfn |= MIGRATE_PFN_WRITE;
		} else {
2273
			if (!(migrate->flags & MIGRATE_VMA_SELECT_SYSTEM))
2274
				goto next;
2275
			pfn = pte_pfn(pte);
2276 2277 2278 2279 2280
			if (is_zero_pfn(pfn)) {
				mpfn = MIGRATE_PFN_MIGRATE;
				migrate->cpages++;
				goto next;
			}
2281
			page = vm_normal_page(migrate->vma, addr, pte);
2282 2283 2284 2285
			mpfn = migrate_pfn(pfn) | MIGRATE_PFN_MIGRATE;
			mpfn |= pte_write(pte) ? MIGRATE_PFN_WRITE : 0;
		}

2286 2287
		/* FIXME support THP */
		if (!page || !page->mapping || PageTransCompound(page)) {
2288
			mpfn = 0;
2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302
			goto next;
		}

		/*
		 * By getting a reference on the page we pin it and that blocks
		 * any kind of migration. Side effect is that it "freezes" the
		 * pte.
		 *
		 * We drop this reference after isolating the page from the lru
		 * for non device page (device page are not on the lru and thus
		 * can't be dropped from it).
		 */
		get_page(page);

2303 2304 2305 2306 2307 2308 2309 2310
		/*
		 * Optimize for the common case where page is only mapped once
		 * in one process. If we can lock the page, then we can safely
		 * set up a special migration page table entry now.
		 */
		if (trylock_page(page)) {
			pte_t swp_pte;

2311
			migrate->cpages++;
2312 2313 2314
			ptep_get_and_clear(mm, addr, ptep);

			/* Setup special migration page table entry */
2315 2316 2317 2318 2319 2320
			if (mpfn & MIGRATE_PFN_WRITE)
				entry = make_writable_migration_entry(
							page_to_pfn(page));
			else
				entry = make_readable_migration_entry(
							page_to_pfn(page));
2321
			swp_pte = swp_entry_to_pte(entry);
2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332
			if (pte_present(pte)) {
				if (pte_soft_dirty(pte))
					swp_pte = pte_swp_mksoft_dirty(swp_pte);
				if (pte_uffd_wp(pte))
					swp_pte = pte_swp_mkuffd_wp(swp_pte);
			} else {
				if (pte_swp_soft_dirty(pte))
					swp_pte = pte_swp_mksoft_dirty(swp_pte);
				if (pte_swp_uffd_wp(pte))
					swp_pte = pte_swp_mkuffd_wp(swp_pte);
			}
2333 2334 2335 2336 2337 2338 2339 2340 2341
			set_pte_at(mm, addr, ptep, swp_pte);

			/*
			 * This is like regular unmap: we remove the rmap and
			 * drop page refcount. Page won't be freed, as we took
			 * a reference just above.
			 */
			page_remove_rmap(page, false);
			put_page(page);
2342 2343 2344

			if (pte_present(pte))
				unmapped++;
2345 2346 2347
		} else {
			put_page(page);
			mpfn = 0;
2348 2349
		}

2350
next:
2351
		migrate->dst[migrate->npages] = 0;
2352 2353
		migrate->src[migrate->npages++] = mpfn;
	}
2354
	arch_leave_lazy_mmu_mode();
2355 2356
	pte_unmap_unlock(ptep - 1, ptl);

2357 2358 2359 2360
	/* Only flush the TLB if we actually modified any entries */
	if (unmapped)
		flush_tlb_range(walk->vma, start, end);

2361 2362 2363
	return 0;
}

2364 2365 2366 2367 2368
static const struct mm_walk_ops migrate_vma_walk_ops = {
	.pmd_entry		= migrate_vma_collect_pmd,
	.pte_hole		= migrate_vma_collect_hole,
};

2369 2370 2371 2372 2373 2374 2375 2376 2377 2378
/*
 * migrate_vma_collect() - collect pages over a range of virtual addresses
 * @migrate: migrate struct containing all migration information
 *
 * This will walk the CPU page table. For each virtual address backed by a
 * valid page, it updates the src array and takes a reference on the page, in
 * order to pin the page until we lock it and unmap it.
 */
static void migrate_vma_collect(struct migrate_vma *migrate)
{
2379
	struct mmu_notifier_range range;
2380

2381 2382 2383 2384 2385
	/*
	 * Note that the pgmap_owner is passed to the mmu notifier callback so
	 * that the registered device driver can skip invalidating device
	 * private page mappings that won't be migrated.
	 */
2386 2387
	mmu_notifier_range_init_owner(&range, MMU_NOTIFY_MIGRATE, 0,
		migrate->vma, migrate->vma->vm_mm, migrate->start, migrate->end,
2388
		migrate->pgmap_owner);
2389
	mmu_notifier_invalidate_range_start(&range);
2390

2391 2392 2393 2394
	walk_page_range(migrate->vma->vm_mm, migrate->start, migrate->end,
			&migrate_vma_walk_ops, migrate);

	mmu_notifier_invalidate_range_end(&range);
2395 2396 2397 2398 2399 2400 2401 2402
	migrate->end = migrate->start + (migrate->npages << PAGE_SHIFT);
}

/*
 * migrate_vma_check_page() - check if page is pinned or not
 * @page: struct page to check
 *
 * Pinned pages cannot be migrated. This is the same test as in
2403
 * folio_migrate_mapping(), except that here we allow migration of a
2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422
 * ZONE_DEVICE page.
 */
static bool migrate_vma_check_page(struct page *page)
{
	/*
	 * One extra ref because caller holds an extra reference, either from
	 * isolate_lru_page() for a regular page, or migrate_vma_collect() for
	 * a device page.
	 */
	int extra = 1;

	/*
	 * FIXME support THP (transparent huge page), it is bit more complex to
	 * check them than regular pages, because they can be mapped with a pmd
	 * or with a pte (split pte mapping).
	 */
	if (PageCompound(page))
		return false;

2423
	/* Page from ZONE_DEVICE have one extra reference */
2424 2425
	if (is_zone_device_page(page))
		extra++;
2426

2427 2428 2429 2430
	/* For file back page */
	if (page_mapping(page))
		extra += 1 + page_has_private(page);

2431 2432 2433 2434 2435 2436 2437
	if ((page_count(page) - extra) > page_mapcount(page))
		return false;

	return true;
}

/*
2438
 * migrate_vma_unmap() - replace page mapping with special migration pte entry
2439 2440
 * @migrate: migrate struct containing all migration information
 *
2441 2442 2443 2444 2445 2446
 * Isolate pages from the LRU and replace mappings (CPU page table pte) with a
 * special migration pte entry and check if it has been pinned. Pinned pages are
 * restored because we cannot migrate them.
 *
 * This is the last step before we call the device driver callback to allocate
 * destination memory and copy contents of original page over to new page.
2447
 */
2448
static void migrate_vma_unmap(struct migrate_vma *migrate)
2449 2450
{
	const unsigned long npages = migrate->npages;
2451
	unsigned long i, restore = 0;
2452 2453 2454 2455
	bool allow_drain = true;

	lru_add_drain();

2456
	for (i = 0; i < npages; i++) {
2457 2458 2459 2460 2461
		struct page *page = migrate_pfn_to_page(migrate->src[i]);

		if (!page)
			continue;

2462 2463 2464 2465 2466 2467 2468
		/* ZONE_DEVICE pages are not on LRU */
		if (!is_zone_device_page(page)) {
			if (!PageLRU(page) && allow_drain) {
				/* Drain CPU's pagevec */
				lru_add_drain_all();
				allow_drain = false;
			}
2469

2470
			if (isolate_lru_page(page)) {
2471 2472 2473
				migrate->src[i] &= ~MIGRATE_PFN_MIGRATE;
				migrate->cpages--;
				restore++;
2474
				continue;
2475
			}
2476 2477 2478

			/* Drop the reference we took in collect */
			put_page(page);
2479 2480
		}

2481 2482
		if (page_mapped(page))
			try_to_migrate(page, 0);
2483

2484 2485 2486 2487
		if (page_mapped(page) || !migrate_vma_check_page(page)) {
			if (!is_zone_device_page(page)) {
				get_page(page);
				putback_lru_page(page);
2488 2489
			}

2490 2491 2492
			migrate->src[i] &= ~MIGRATE_PFN_MIGRATE;
			migrate->cpages--;
			restore++;
2493
			continue;
2494 2495 2496
		}
	}

2497
	for (i = 0; i < npages && restore; i++) {
2498 2499 2500 2501 2502 2503 2504 2505 2506
		struct page *page = migrate_pfn_to_page(migrate->src[i]);

		if (!page || (migrate->src[i] & MIGRATE_PFN_MIGRATE))
			continue;

		remove_migration_ptes(page, page, false);

		migrate->src[i] = 0;
		unlock_page(page);
2507
		put_page(page);
2508 2509 2510 2511
		restore--;
	}
}

2512 2513
/**
 * migrate_vma_setup() - prepare to migrate a range of memory
2514
 * @args: contains the vma, start, and pfns arrays for the migration
2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530
 *
 * Returns: negative errno on failures, 0 when 0 or more pages were migrated
 * without an error.
 *
 * Prepare to migrate a range of memory virtual address range by collecting all
 * the pages backing each virtual address in the range, saving them inside the
 * src array.  Then lock those pages and unmap them. Once the pages are locked
 * and unmapped, check whether each page is pinned or not.  Pages that aren't
 * pinned have the MIGRATE_PFN_MIGRATE flag set (by this function) in the
 * corresponding src array entry.  Then restores any pages that are pinned, by
 * remapping and unlocking those pages.
 *
 * The caller should then allocate destination memory and copy source memory to
 * it for all those entries (ie with MIGRATE_PFN_VALID and MIGRATE_PFN_MIGRATE
 * flag set).  Once these are allocated and copied, the caller must update each
 * corresponding entry in the dst array with the pfn value of the destination
2531 2532
 * page and with MIGRATE_PFN_VALID. Destination pages must be locked via
 * lock_page().
2533 2534 2535 2536 2537 2538 2539 2540 2541 2542
 *
 * Note that the caller does not have to migrate all the pages that are marked
 * with MIGRATE_PFN_MIGRATE flag in src array unless this is a migration from
 * device memory to system memory.  If the caller cannot migrate a device page
 * back to system memory, then it must return VM_FAULT_SIGBUS, which has severe
 * consequences for the userspace process, so it must be avoided if at all
 * possible.
 *
 * For empty entries inside CPU page table (pte_none() or pmd_none() is true) we
 * do set MIGRATE_PFN_MIGRATE flag inside the corresponding source array thus
I
Ingo Molnar 已提交
2543 2544
 * allowing the caller to allocate device memory for those unbacked virtual
 * addresses.  For this the caller simply has to allocate device memory and
2545
 * properly set the destination entry like for regular migration.  Note that
I
Ingo Molnar 已提交
2546 2547
 * this can still fail, and thus inside the device driver you must check if the
 * migration was successful for those entries after calling migrate_vma_pages(),
2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566
 * just like for regular migration.
 *
 * After that, the callers must call migrate_vma_pages() to go over each entry
 * in the src array that has the MIGRATE_PFN_VALID and MIGRATE_PFN_MIGRATE flag
 * set. If the corresponding entry in dst array has MIGRATE_PFN_VALID flag set,
 * then migrate_vma_pages() to migrate struct page information from the source
 * struct page to the destination struct page.  If it fails to migrate the
 * struct page information, then it clears the MIGRATE_PFN_MIGRATE flag in the
 * src array.
 *
 * At this point all successfully migrated pages have an entry in the src
 * array with MIGRATE_PFN_VALID and MIGRATE_PFN_MIGRATE flag set and the dst
 * array entry with MIGRATE_PFN_VALID flag set.
 *
 * Once migrate_vma_pages() returns the caller may inspect which pages were
 * successfully migrated, and which were not.  Successfully migrated pages will
 * have the MIGRATE_PFN_MIGRATE flag set for their src array entry.
 *
 * It is safe to update device page table after migrate_vma_pages() because
2567
 * both destination and source page are still locked, and the mmap_lock is held
2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613
 * in read mode (hence no one can unmap the range being migrated).
 *
 * Once the caller is done cleaning up things and updating its page table (if it
 * chose to do so, this is not an obligation) it finally calls
 * migrate_vma_finalize() to update the CPU page table to point to new pages
 * for successfully migrated pages or otherwise restore the CPU page table to
 * point to the original source pages.
 */
int migrate_vma_setup(struct migrate_vma *args)
{
	long nr_pages = (args->end - args->start) >> PAGE_SHIFT;

	args->start &= PAGE_MASK;
	args->end &= PAGE_MASK;
	if (!args->vma || is_vm_hugetlb_page(args->vma) ||
	    (args->vma->vm_flags & VM_SPECIAL) || vma_is_dax(args->vma))
		return -EINVAL;
	if (nr_pages <= 0)
		return -EINVAL;
	if (args->start < args->vma->vm_start ||
	    args->start >= args->vma->vm_end)
		return -EINVAL;
	if (args->end <= args->vma->vm_start || args->end > args->vma->vm_end)
		return -EINVAL;
	if (!args->src || !args->dst)
		return -EINVAL;

	memset(args->src, 0, sizeof(*args->src) * nr_pages);
	args->cpages = 0;
	args->npages = 0;

	migrate_vma_collect(args);

	if (args->cpages)
		migrate_vma_unmap(args);

	/*
	 * At this point pages are locked and unmapped, and thus they have
	 * stable content and can safely be copied to destination memory that
	 * is allocated by the drivers.
	 */
	return 0;

}
EXPORT_SYMBOL(migrate_vma_setup);

2614 2615 2616 2617 2618 2619 2620 2621
/*
 * This code closely matches the code in:
 *   __handle_mm_fault()
 *     handle_pte_fault()
 *       do_anonymous_page()
 * to map in an anonymous zero page but the struct page will be a ZONE_DEVICE
 * private page.
 */
2622 2623 2624
static void migrate_vma_insert_page(struct migrate_vma *migrate,
				    unsigned long addr,
				    struct page *page,
2625
				    unsigned long *src)
2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660
{
	struct vm_area_struct *vma = migrate->vma;
	struct mm_struct *mm = vma->vm_mm;
	bool flush = false;
	spinlock_t *ptl;
	pte_t entry;
	pgd_t *pgdp;
	p4d_t *p4dp;
	pud_t *pudp;
	pmd_t *pmdp;
	pte_t *ptep;

	/* Only allow populating anonymous memory */
	if (!vma_is_anonymous(vma))
		goto abort;

	pgdp = pgd_offset(mm, addr);
	p4dp = p4d_alloc(mm, pgdp, addr);
	if (!p4dp)
		goto abort;
	pudp = pud_alloc(mm, p4dp, addr);
	if (!pudp)
		goto abort;
	pmdp = pmd_alloc(mm, pudp, addr);
	if (!pmdp)
		goto abort;

	if (pmd_trans_huge(*pmdp) || pmd_devmap(*pmdp))
		goto abort;

	/*
	 * Use pte_alloc() instead of pte_alloc_map().  We can't run
	 * pte_offset_map() on pmds where a huge pmd might be created
	 * from a different thread.
	 *
2661
	 * pte_alloc_map() is safe to use under mmap_write_lock(mm) or when
2662 2663
	 * parallel threads are excluded by other means.
	 *
2664
	 * Here we only have mmap_read_lock(mm).
2665
	 */
2666
	if (pte_alloc(mm, pmdp))
2667 2668 2669 2670 2671 2672 2673 2674
		goto abort;

	/* See the comment in pte_alloc_one_map() */
	if (unlikely(pmd_trans_unstable(pmdp)))
		goto abort;

	if (unlikely(anon_vma_prepare(vma)))
		goto abort;
2675
	if (mem_cgroup_charge(page_folio(page), vma->vm_mm, GFP_KERNEL))
2676 2677 2678 2679 2680 2681 2682 2683 2684
		goto abort;

	/*
	 * The memory barrier inside __SetPageUptodate makes sure that
	 * preceding stores to the page contents become visible before
	 * the set_pte_at() write.
	 */
	__SetPageUptodate(page);

2685 2686 2687 2688
	if (is_zone_device_page(page)) {
		if (is_device_private_page(page)) {
			swp_entry_t swp_entry;

2689 2690 2691 2692 2693 2694
			if (vma->vm_flags & VM_WRITE)
				swp_entry = make_writable_device_private_entry(
							page_to_pfn(page));
			else
				swp_entry = make_readable_device_private_entry(
							page_to_pfn(page));
2695
			entry = swp_entry_to_pte(swp_entry);
2696 2697 2698 2699 2700 2701 2702
		} else {
			/*
			 * For now we only support migrating to un-addressable
			 * device memory.
			 */
			pr_warn_once("Unsupported ZONE_DEVICE page type.\n");
			goto abort;
2703
		}
2704 2705 2706 2707 2708 2709 2710 2711
	} else {
		entry = mk_pte(page, vma->vm_page_prot);
		if (vma->vm_flags & VM_WRITE)
			entry = pte_mkwrite(pte_mkdirty(entry));
	}

	ptep = pte_offset_map_lock(mm, pmdp, addr, &ptl);

2712 2713 2714
	if (check_stable_address_space(mm))
		goto unlock_abort;

2715 2716 2717
	if (pte_present(*ptep)) {
		unsigned long pfn = pte_pfn(*ptep);

2718 2719
		if (!is_zero_pfn(pfn))
			goto unlock_abort;
2720
		flush = true;
2721 2722
	} else if (!pte_none(*ptep))
		goto unlock_abort;
2723 2724

	/*
2725
	 * Check for userfaultfd but do not deliver the fault. Instead,
2726 2727
	 * just back off.
	 */
2728 2729
	if (userfaultfd_missing(vma))
		goto unlock_abort;
2730 2731

	inc_mm_counter(mm, MM_ANONPAGES);
2732
	page_add_new_anon_rmap(page, vma, addr, false);
2733
	if (!is_zone_device_page(page))
2734
		lru_cache_add_inactive_or_unevictable(page, vma);
2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751
	get_page(page);

	if (flush) {
		flush_cache_page(vma, addr, pte_pfn(*ptep));
		ptep_clear_flush_notify(vma, addr, ptep);
		set_pte_at_notify(mm, addr, ptep, entry);
		update_mmu_cache(vma, addr, ptep);
	} else {
		/* No need to invalidate - it was non-present before */
		set_pte_at(mm, addr, ptep, entry);
		update_mmu_cache(vma, addr, ptep);
	}

	pte_unmap_unlock(ptep, ptl);
	*src = MIGRATE_PFN_MIGRATE;
	return;

2752 2753
unlock_abort:
	pte_unmap_unlock(ptep, ptl);
2754 2755 2756 2757
abort:
	*src &= ~MIGRATE_PFN_MIGRATE;
}

2758
/**
2759 2760 2761 2762 2763 2764 2765
 * migrate_vma_pages() - migrate meta-data from src page to dst page
 * @migrate: migrate struct containing all migration information
 *
 * This migrates struct page meta-data from source struct page to destination
 * struct page. This effectively finishes the migration from source page to the
 * destination page.
 */
2766
void migrate_vma_pages(struct migrate_vma *migrate)
2767 2768 2769
{
	const unsigned long npages = migrate->npages;
	const unsigned long start = migrate->start;
2770 2771
	struct mmu_notifier_range range;
	unsigned long addr, i;
2772
	bool notified = false;
2773 2774 2775 2776 2777 2778 2779

	for (i = 0, addr = start; i < npages; addr += PAGE_SIZE, i++) {
		struct page *newpage = migrate_pfn_to_page(migrate->dst[i]);
		struct page *page = migrate_pfn_to_page(migrate->src[i]);
		struct address_space *mapping;
		int r;

2780 2781
		if (!newpage) {
			migrate->src[i] &= ~MIGRATE_PFN_MIGRATE;
2782
			continue;
2783 2784 2785
		}

		if (!page) {
2786
			if (!(migrate->src[i] & MIGRATE_PFN_MIGRATE))
2787 2788 2789
				continue;
			if (!notified) {
				notified = true;
2790

2791 2792 2793
				mmu_notifier_range_init_owner(&range,
					MMU_NOTIFY_MIGRATE, 0, migrate->vma,
					migrate->vma->vm_mm, addr, migrate->end,
2794
					migrate->pgmap_owner);
2795
				mmu_notifier_invalidate_range_start(&range);
2796 2797
			}
			migrate_vma_insert_page(migrate, addr, newpage,
2798
						&migrate->src[i]);
2799
			continue;
2800
		}
2801 2802 2803

		mapping = page_mapping(page);

2804 2805 2806 2807 2808 2809 2810 2811 2812 2813
		if (is_zone_device_page(newpage)) {
			if (is_device_private_page(newpage)) {
				/*
				 * For now only support private anonymous when
				 * migrating to un-addressable device memory.
				 */
				if (mapping) {
					migrate->src[i] &= ~MIGRATE_PFN_MIGRATE;
					continue;
				}
2814
			} else {
2815 2816 2817 2818 2819 2820 2821 2822 2823
				/*
				 * Other types of ZONE_DEVICE page are not
				 * supported.
				 */
				migrate->src[i] &= ~MIGRATE_PFN_MIGRATE;
				continue;
			}
		}

2824 2825 2826 2827
		r = migrate_page(mapping, newpage, page, MIGRATE_SYNC_NO_COPY);
		if (r != MIGRATEPAGE_SUCCESS)
			migrate->src[i] &= ~MIGRATE_PFN_MIGRATE;
	}
2828

2829 2830 2831 2832 2833
	/*
	 * No need to double call mmu_notifier->invalidate_range() callback as
	 * the above ptep_clear_flush_notify() inside migrate_vma_insert_page()
	 * did already call it.
	 */
2834
	if (notified)
2835
		mmu_notifier_invalidate_range_only_end(&range);
2836
}
2837
EXPORT_SYMBOL(migrate_vma_pages);
2838

2839
/**
2840 2841 2842 2843 2844 2845 2846 2847 2848 2849
 * migrate_vma_finalize() - restore CPU page table entry
 * @migrate: migrate struct containing all migration information
 *
 * This replaces the special migration pte entry with either a mapping to the
 * new page if migration was successful for that page, or to the original page
 * otherwise.
 *
 * This also unlocks the pages and puts them back on the lru, or drops the extra
 * refcount, for device pages.
 */
2850
void migrate_vma_finalize(struct migrate_vma *migrate)
2851 2852 2853 2854 2855 2856 2857 2858
{
	const unsigned long npages = migrate->npages;
	unsigned long i;

	for (i = 0; i < npages; i++) {
		struct page *newpage = migrate_pfn_to_page(migrate->dst[i]);
		struct page *page = migrate_pfn_to_page(migrate->src[i]);

2859 2860 2861 2862 2863
		if (!page) {
			if (newpage) {
				unlock_page(newpage);
				put_page(newpage);
			}
2864
			continue;
2865 2866
		}

2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877
		if (!(migrate->src[i] & MIGRATE_PFN_MIGRATE) || !newpage) {
			if (newpage) {
				unlock_page(newpage);
				put_page(newpage);
			}
			newpage = page;
		}

		remove_migration_ptes(page, newpage, false);
		unlock_page(page);

2878 2879 2880 2881
		if (is_zone_device_page(page))
			put_page(page);
		else
			putback_lru_page(page);
2882 2883 2884

		if (newpage != page) {
			unlock_page(newpage);
2885 2886 2887 2888
			if (is_zone_device_page(newpage))
				put_page(newpage);
			else
				putback_lru_page(newpage);
2889 2890 2891
		}
	}
}
2892
EXPORT_SYMBOL(migrate_vma_finalize);
2893
#endif /* CONFIG_DEVICE_PRIVATE */
2894

2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026
/*
 * node_demotion[] example:
 *
 * Consider a system with two sockets.  Each socket has
 * three classes of memory attached: fast, medium and slow.
 * Each memory class is placed in its own NUMA node.  The
 * CPUs are placed in the node with the "fast" memory.  The
 * 6 NUMA nodes (0-5) might be split among the sockets like
 * this:
 *
 *	Socket A: 0, 1, 2
 *	Socket B: 3, 4, 5
 *
 * When Node 0 fills up, its memory should be migrated to
 * Node 1.  When Node 1 fills up, it should be migrated to
 * Node 2.  The migration path start on the nodes with the
 * processors (since allocations default to this node) and
 * fast memory, progress through medium and end with the
 * slow memory:
 *
 *	0 -> 1 -> 2 -> stop
 *	3 -> 4 -> 5 -> stop
 *
 * This is represented in the node_demotion[] like this:
 *
 *	{  nr=1, nodes[0]=1 }, // Node 0 migrates to 1
 *	{  nr=1, nodes[0]=2 }, // Node 1 migrates to 2
 *	{  nr=0, nodes[0]=-1 }, // Node 2 does not migrate
 *	{  nr=1, nodes[0]=4 }, // Node 3 migrates to 4
 *	{  nr=1, nodes[0]=5 }, // Node 4 migrates to 5
 *	{  nr=0, nodes[0]=-1 }, // Node 5 does not migrate
 *
 * Moreover some systems may have multiple slow memory nodes.
 * Suppose a system has one socket with 3 memory nodes, node 0
 * is fast memory type, and node 1/2 both are slow memory
 * type, and the distance between fast memory node and slow
 * memory node is same. So the migration path should be:
 *
 *	0 -> 1/2 -> stop
 *
 * This is represented in the node_demotion[] like this:
 *	{ nr=2, {nodes[0]=1, nodes[1]=2} }, // Node 0 migrates to node 1 and node 2
 *	{ nr=0, nodes[0]=-1, }, // Node 1 dose not migrate
 *	{ nr=0, nodes[0]=-1, }, // Node 2 does not migrate
 */

/*
 * Writes to this array occur without locking.  Cycles are
 * not allowed: Node X demotes to Y which demotes to X...
 *
 * If multiple reads are performed, a single rcu_read_lock()
 * must be held over all reads to ensure that no cycles are
 * observed.
 */
#define DEFAULT_DEMOTION_TARGET_NODES 15

#if MAX_NUMNODES < DEFAULT_DEMOTION_TARGET_NODES
#define DEMOTION_TARGET_NODES	(MAX_NUMNODES - 1)
#else
#define DEMOTION_TARGET_NODES	DEFAULT_DEMOTION_TARGET_NODES
#endif

struct demotion_nodes {
	unsigned short nr;
	short nodes[DEMOTION_TARGET_NODES];
};

static struct demotion_nodes *node_demotion __read_mostly;

/**
 * next_demotion_node() - Get the next node in the demotion path
 * @node: The starting node to lookup the next node
 *
 * Return: node id for next memory node in the demotion path hierarchy
 * from @node; NUMA_NO_NODE if @node is terminal.  This does not keep
 * @node online or guarantee that it *continues* to be the next demotion
 * target.
 */
int next_demotion_node(int node)
{
	struct demotion_nodes *nd;
	unsigned short target_nr, index;
	int target;

	if (!node_demotion)
		return NUMA_NO_NODE;

	nd = &node_demotion[node];

	/*
	 * node_demotion[] is updated without excluding this
	 * function from running.  RCU doesn't provide any
	 * compiler barriers, so the READ_ONCE() is required
	 * to avoid compiler reordering or read merging.
	 *
	 * Make sure to use RCU over entire code blocks if
	 * node_demotion[] reads need to be consistent.
	 */
	rcu_read_lock();
	target_nr = READ_ONCE(nd->nr);

	switch (target_nr) {
	case 0:
		target = NUMA_NO_NODE;
		goto out;
	case 1:
		index = 0;
		break;
	default:
		/*
		 * If there are multiple target nodes, just select one
		 * target node randomly.
		 *
		 * In addition, we can also use round-robin to select
		 * target node, but we should introduce another variable
		 * for node_demotion[] to record last selected target node,
		 * that may cause cache ping-pong due to the changing of
		 * last target node. Or introducing per-cpu data to avoid
		 * caching issue, which seems more complicated. So selecting
		 * target node randomly seems better until now.
		 */
		index = get_random_int() % target_nr;
		break;
	}

	target = READ_ONCE(nd->nodes[index]);

out:
	rcu_read_unlock();
	return target;
}

3027
#if defined(CONFIG_HOTPLUG_CPU)
3028 3029 3030
/* Disable reclaim-based migration. */
static void __disable_all_migrate_targets(void)
{
3031
	int node, i;
3032

3033 3034
	if (!node_demotion)
		return;
3035

3036 3037 3038 3039 3040
	for_each_online_node(node) {
		node_demotion[node].nr = 0;
		for (i = 0; i < DEMOTION_TARGET_NODES; i++)
			node_demotion[node].nodes[i] = NUMA_NO_NODE;
	}
3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066
}

static void disable_all_migrate_targets(void)
{
	__disable_all_migrate_targets();

	/*
	 * Ensure that the "disable" is visible across the system.
	 * Readers will see either a combination of before+disable
	 * state or disable+after.  They will never see before and
	 * after state together.
	 *
	 * The before+after state together might have cycles and
	 * could cause readers to do things like loop until this
	 * function finishes.  This ensures they can only see a
	 * single "bad" read and would, for instance, only loop
	 * once.
	 */
	synchronize_rcu();
}

/*
 * Find an automatic demotion target for 'node'.
 * Failing here is OK.  It might just indicate
 * being at the end of a chain.
 */
3067 3068
static int establish_migrate_target(int node, nodemask_t *used,
				    int best_distance)
3069
{
3070 3071
	int migration_target, index, val;
	struct demotion_nodes *nd;
3072

3073
	if (!node_demotion)
3074 3075
		return NUMA_NO_NODE;

3076 3077
	nd = &node_demotion[node];

3078 3079 3080 3081
	migration_target = find_next_best_node(node, used);
	if (migration_target == NUMA_NO_NODE)
		return NUMA_NO_NODE;

3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100
	/*
	 * If the node has been set a migration target node before,
	 * which means it's the best distance between them. Still
	 * check if this node can be demoted to other target nodes
	 * if they have a same best distance.
	 */
	if (best_distance != -1) {
		val = node_distance(node, migration_target);
		if (val > best_distance)
			return NUMA_NO_NODE;
	}

	index = nd->nr;
	if (WARN_ONCE(index >= DEMOTION_TARGET_NODES,
		      "Exceeds maximum demotion target nodes\n"))
		return NUMA_NO_NODE;

	nd->nodes[index] = migration_target;
	nd->nr++;
3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115

	return migration_target;
}

/*
 * When memory fills up on a node, memory contents can be
 * automatically migrated to another node instead of
 * discarded at reclaim.
 *
 * Establish a "migration path" which will start at nodes
 * with CPUs and will follow the priorities used to build the
 * page allocator zonelists.
 *
 * The difference here is that cycles must be avoided.  If
 * node0 migrates to node1, then neither node1, nor anything
3116 3117 3118
 * node1 migrates to can migrate to node0. Also one node can
 * be migrated to multiple nodes if the target nodes all have
 * a same best-distance against the source node.
3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129
 *
 * This function can run simultaneously with readers of
 * node_demotion[].  However, it can not run simultaneously
 * with itself.  Exclusion is provided by memory hotplug events
 * being single-threaded.
 */
static void __set_migration_target_nodes(void)
{
	nodemask_t next_pass	= NODE_MASK_NONE;
	nodemask_t this_pass	= NODE_MASK_NONE;
	nodemask_t used_targets = NODE_MASK_NONE;
3130
	int node, best_distance;
3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159

	/*
	 * Avoid any oddities like cycles that could occur
	 * from changes in the topology.  This will leave
	 * a momentary gap when migration is disabled.
	 */
	disable_all_migrate_targets();

	/*
	 * Allocations go close to CPUs, first.  Assume that
	 * the migration path starts at the nodes with CPUs.
	 */
	next_pass = node_states[N_CPU];
again:
	this_pass = next_pass;
	next_pass = NODE_MASK_NONE;
	/*
	 * To avoid cycles in the migration "graph", ensure
	 * that migration sources are not future targets by
	 * setting them in 'used_targets'.  Do this only
	 * once per pass so that multiple source nodes can
	 * share a target node.
	 *
	 * 'used_targets' will become unavailable in future
	 * passes.  This limits some opportunities for
	 * multiple source nodes to share a destination.
	 */
	nodes_or(used_targets, used_targets, this_pass);

3160 3161
	for_each_node_mask(node, this_pass) {
		best_distance = -1;
3162 3163

		/*
3164 3165 3166
		 * Try to set up the migration path for the node, and the target
		 * migration nodes can be multiple, so doing a loop to find all
		 * the target nodes if they all have a best node distance.
3167
		 */
3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185
		do {
			int target_node =
				establish_migrate_target(node, &used_targets,
							 best_distance);

			if (target_node == NUMA_NO_NODE)
				break;

			if (best_distance == -1)
				best_distance = node_distance(node, target_node);

			/*
			 * Visit targets from this pass in the next pass.
			 * Eventually, every node will have been part of
			 * a pass, and will become set in 'used_targets'.
			 */
			node_set(target_node, next_pass);
		} while (1);
3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204
	}
	/*
	 * 'next_pass' contains nodes which became migration
	 * targets in this pass.  Make additional passes until
	 * no more migrations targets are available.
	 */
	if (!nodes_empty(next_pass))
		goto again;
}

/*
 * For callers that do not hold get_online_mems() already.
 */
static void set_migration_target_nodes(void)
{
	get_online_mems();
	__set_migration_target_nodes();
	put_online_mems();
}
3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217

/*
 * This leaves migrate-on-reclaim transiently disabled between
 * the MEM_GOING_OFFLINE and MEM_OFFLINE events.  This runs
 * whether reclaim-based migration is enabled or not, which
 * ensures that the user can turn reclaim-based migration at
 * any time without needing to recalculate migration targets.
 *
 * These callbacks already hold get_online_mems().  That is why
 * __set_migration_target_nodes() can be used as opposed to
 * set_migration_target_nodes().
 */
static int __meminit migrate_on_reclaim_callback(struct notifier_block *self,
3218
						 unsigned long action, void *_arg)
3219
{
3220 3221 3222 3223 3224 3225 3226 3227 3228 3229
	struct memory_notify *arg = _arg;

	/*
	 * Only update the node migration order when a node is
	 * changing status, like online->offline.  This avoids
	 * the overhead of synchronize_rcu() in most cases.
	 */
	if (arg->status_change_nid < 0)
		return notifier_from_errno(0);

3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262
	switch (action) {
	case MEM_GOING_OFFLINE:
		/*
		 * Make sure there are not transient states where
		 * an offline node is a migration target.  This
		 * will leave migration disabled until the offline
		 * completes and the MEM_OFFLINE case below runs.
		 */
		disable_all_migrate_targets();
		break;
	case MEM_OFFLINE:
	case MEM_ONLINE:
		/*
		 * Recalculate the target nodes once the node
		 * reaches its final state (online or offline).
		 */
		__set_migration_target_nodes();
		break;
	case MEM_CANCEL_OFFLINE:
		/*
		 * MEM_GOING_OFFLINE disabled all the migration
		 * targets.  Reenable them.
		 */
		__set_migration_target_nodes();
		break;
	case MEM_GOING_ONLINE:
	case MEM_CANCEL_ONLINE:
		break;
	}

	return notifier_from_errno(0);
}

3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281
/*
 * React to hotplug events that might affect the migration targets
 * like events that online or offline NUMA nodes.
 *
 * The ordering is also currently dependent on which nodes have
 * CPUs.  That means we need CPU on/offline notification too.
 */
static int migration_online_cpu(unsigned int cpu)
{
	set_migration_target_nodes();
	return 0;
}

static int migration_offline_cpu(unsigned int cpu)
{
	set_migration_target_nodes();
	return 0;
}

3282 3283 3284 3285
static int __init migrate_on_reclaim_init(void)
{
	int ret;

3286 3287 3288 3289 3290
	node_demotion = kmalloc_array(nr_node_ids,
				      sizeof(struct demotion_nodes),
				      GFP_KERNEL);
	WARN_ON(!node_demotion);

3291 3292
	ret = cpuhp_setup_state_nocalls(CPUHP_MM_DEMOTION_DEAD, "mm/demotion:offline",
					NULL, migration_offline_cpu);
3293 3294 3295 3296 3297 3298 3299
	/*
	 * In the unlikely case that this fails, the automatic
	 * migration targets may become suboptimal for nodes
	 * where N_CPU changes.  With such a small impact in a
	 * rare case, do not bother trying to do anything special.
	 */
	WARN_ON(ret < 0);
3300 3301 3302
	ret = cpuhp_setup_state(CPUHP_AP_MM_DEMOTION_ONLINE, "mm/demotion:online",
				migration_online_cpu, NULL);
	WARN_ON(ret < 0);
3303 3304 3305 3306 3307

	hotplug_memory_notifier(migrate_on_reclaim_callback, 100);
	return 0;
}
late_initcall(migrate_on_reclaim_init);
3308
#endif /* CONFIG_HOTPLUG_CPU */
3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369

bool numa_demotion_enabled = false;

#ifdef CONFIG_SYSFS
static ssize_t numa_demotion_enabled_show(struct kobject *kobj,
					  struct kobj_attribute *attr, char *buf)
{
	return sysfs_emit(buf, "%s\n",
			  numa_demotion_enabled ? "true" : "false");
}

static ssize_t numa_demotion_enabled_store(struct kobject *kobj,
					   struct kobj_attribute *attr,
					   const char *buf, size_t count)
{
	if (!strncmp(buf, "true", 4) || !strncmp(buf, "1", 1))
		numa_demotion_enabled = true;
	else if (!strncmp(buf, "false", 5) || !strncmp(buf, "0", 1))
		numa_demotion_enabled = false;
	else
		return -EINVAL;

	return count;
}

static struct kobj_attribute numa_demotion_enabled_attr =
	__ATTR(demotion_enabled, 0644, numa_demotion_enabled_show,
	       numa_demotion_enabled_store);

static struct attribute *numa_attrs[] = {
	&numa_demotion_enabled_attr.attr,
	NULL,
};

static const struct attribute_group numa_attr_group = {
	.attrs = numa_attrs,
};

static int __init numa_init_sysfs(void)
{
	int err;
	struct kobject *numa_kobj;

	numa_kobj = kobject_create_and_add("numa", mm_kobj);
	if (!numa_kobj) {
		pr_err("failed to create numa kobject\n");
		return -ENOMEM;
	}
	err = sysfs_create_group(numa_kobj, &numa_attr_group);
	if (err) {
		pr_err("failed to register numa group\n");
		goto delete_obj;
	}
	return 0;

delete_obj:
	kobject_put(numa_kobj);
	return err;
}
subsys_initcall(numa_init_sysfs);
#endif
反馈
建议
客服 返回
顶部